You Cannot Outrun the Global Economy

The success of a globally diversified fund is tied to the health of the entire world economy. If the global economy suffers from climate change, the fund suffers with it. The fund now treats climate risk as a core investment risk, placing it on the same level as market risk, credit risk, and counterparty risk.

Many people think climate action is a choice between profit and the planet. For a long-term investor, this is a false choice. Research shows that climate change is a material financial risk that can destroy the value of investments. Because the fund is so large and diverse, it cannot simply hide from these shifts.

“The global economy cannot outrun climate change, so neither can our investments.” — Nicolai Tangen, CEO of Norges Bank Investment Management

Talking to Companies Works Better Than Walking Away

When the fund identifies a company with high emissions, it has two choices. It can sell its shares, which is called divestment, or it can stay and push for change. The fund prefers the second path, known as engagement-led action.

The fund’s data shows that this approach gets results. Companies that have direct, structured dialogues with the fund are more likely to set net-zero targets than those that do not.

• 17 percent of companies the fund engaged with set net-zero targets.

• 16 percent of high-emitting peers set targets.

• Only 2 to 7 percent of companies in other peer groups set similar targets.

Selling shares often just passes the problem to a new owner who may not care about the environment. By staying as an owner, the fund can influence how a company operates and push for real-world changes.

“Divesting doesn’t reduce the amount of carbon in the atmosphere, engagement can.” — Carine Smith Ihenacho, Chief Governance and Compliance Officer

The Cost of Inaction

The financial argument for climate action is backed by alarming data. Physical changes to the climate are already affecting wealth on a global scale. Research cited by the fund shows that global GDP per capita was more than 20 percent lower between 1960 and 2019 than it would have been without physical climate change.

The future risks are even more significant. The fund notes that a temperature rise of just 1°C could impact global GDP by over 20 percent. For a long-term investor, these numbers are a wake-up call. They represent a permanent threat to the stability of global markets and the value of the fund and must not be treated as minor fluctuations.

Nature and Climate are Two Sides of the Same Coin

The fund’s strategy for 2030 has evolved to treat nature loss and climate change as parts of the same problem. Ecosystems like forests, wetlands, and oceans are vital because they store carbon. When these ecosystems are destroyed, that carbon is released and global warming accelerates.

The 2030 plan includes a sharp focus on the need to eliminate deforestation and the conversion of natural ecosystems in the short term. Additionally, the fund aims for “no net loss of nature” for its new investments in renewable energy infrastructure. This shift shows the fund now considers how the destruction of the broader environment creates direct financial risks for its portfolio.

Using Artificial Intelligence to Spot Transition Winners

Managing thousands of companies requires high-tech tools. The fund is now embedding artificial intelligence and proprietary analytics into its strategy. One of the key tools is a set of proprietary climate change expectation scores. These scores allow the fund to benchmark companies and see how well they are prepared for a low-carbon future.

The fund also uses an Investment Simulator. This is a specialised platform for trading analysis used by portfolio managers. By using AI to structure data from thousands of company meetings, managers can identify which industries are leading the way. These tools help the fund identify “transition winners,” which are the businesses best positioned to thrive as the world shifts away from fossil fuels.

Ending Note

The scientific link between greenhouse gas emissions and rising temperatures is certain. However, the economic path forward is much less clear. Government policies around the world have been unpredictable. Policy support for climate action has been promised, delivered, and then sometimes removed in waves.

The fund warns of a “disorderly transition.” This occurs because of policy delay and divergence globally. When countries do not move together or follow a clear path, it creates uncertainty for businesses and investors. As the global economy moves toward net-zero emissions, every investor must face a difficult reality. Are your local policies and your own investments prepared for this inevitable but unpredictable economic shift?

Find NIBM’s Climate Action Plan 2030 here.

The previous installments in this series highlighted that households adopt solar technology or change what they eat when doing so solves a problem within the conditions of their daily lives. Transport raises a related question: what determines whether a household can actually change the way it moves?

Transport policy often treats travel as a choice problem. Make driving more expensive, provide a greener alternative, or give people information about walking and cycling, and people should switch. But everyday travel is embedded in routines, constraints, and social circumstances. The relevant question is therefore whether a greener option is genuinely usable and what would make people change an established way of travelling.

The cases below show four different answers. Sometimes changing the conditions around a routine opens space for people to change it. Sometimes removing one barrier moves a second, related one with it, while others stay fixed. Sometimes motivation cannot overcome a journey that is structurally difficult to change. And sometimes changing the cost of the old option works because a workable alternative is strengthened at the same time.

UK Sustainable Travel Towns

The UK Sustainable Travel Towns programme, run in Darlington, Peterborough, and Worcester between 2004 and 2009, combined personalized travel planning, public transport information and marketing, workplace and school travel schemes, walking and cycling promotion, and bus network improvements, delivered across whole towns at the same time.

By the end of the programme, car trips per person across the three towns had fallen by about 9 percent, against a decline of roughly 1 percent in similar towns nationally. Bus trips rose by 10 to 22 percent, cycling by 26 to 30 percent, and walking by 10 to 13 percent, while all three were falling nationally. In Darlington, which also received separate funding as a Cycling Demonstration Town, cycling trips rose by up to 113 percent (Sloman et al., 2010).

The evaluation reported two separate findings about who changed. The greatest shifts came from college students, people looking for work, and the recently retired. The lowest rates of change came from full- and part-time workers and from intensive car users (Sloman et al., 2010). The evaluation does not explain why these groups differed so much. One plausible reading, consistent with research on habit and context change more generally, is that people whose daily routine was already being rebuilt for other reasons had more room to reconsider how they got around, while people in stable, established jobs had less occasion to revisit a routine that was otherwise unchanged (Verplanken and Wood, 2006; Verplanken, Walker, Davis, and Jurasek, 2008). The programme was not designed to test this interpretation directly; it remains a plausible reading of the pattern rather than a demonstrated finding.

The change was also not simple mode-for-mode substitution. Alongside people switching from car to bus or bicycle, some households changed the destination of a trip, shopping locally rather than travelling to an out-of-town supermarket, and some trips appear to have stopped happening altogether when people found another way to meet the same need without making the journey (Sloman et al., 2010).

Delhi’s Free Bus Scheme for Women

Delhi’s free bus travel scheme for women, introduced in October 2019 alongside the deployment of more than 11,000 security marshals on buses, sat within a broader effort to improve women’s mobility and address safety concerns on public transport. Fare affordability was one of several barriers the policy targeted.

It produced measurable changes in how women travelled. A World Resources Institute India study of more than 2,000 women bus users found that women’s share of daily bus trips rose from about 33 to 42 percent between 2019 and 2023, with the largest gains among lower-income women. Half of those surveyed reported saving at least Rs 500 a month, and many reported greater independence and improved access to work, education, and healthcare (Jamba, Devaraj, and Kanuri, 2025). A separate study using India’s nationally representative Time Use Survey and a matched difference-in-differences design found that women in Delhi became 6.4 percentage points more likely to commute after sunset following the policy, and less likely to travel with a companion (Ratra and Kumar, 2026).

That second finding reveals a change in the practical choice set, beyond a rise in bus use. Travelling after dark and travelling alone are behaviours shaped by safety concerns as much as by cost. The evidence suggests the policy loosened those specific constraints, letting some women travel at times and in circumstances they had previously avoided.

Expanding the choice set in one direction leaves other constraints untouched. The WRI study documents continuing overcrowding, unreliable services, and a shrinking bus fleet. Women travelling longer distances often had to switch to the Metro, where the fare waiver did not apply (Jamba, Devaraj, and Kanuri, 2025). A 2024 study tracing the first-mile trip, the main public transport leg, and the last-mile trip across women from different income groups in Delhi found that the first mile was consistently the weakest link, with low-income women facing the greatest difficulty completing a journey door to door (Sil, Chowdhury, and Thoreau, 2024).

Free bus travel worked: it removed a real barrier and measurably expanded women’s mobility. What the case shows is that usability has several dimensions. Cost determines whether someone can afford a journey. Safety determines whether they are willing to make it. Reliability, route coverage, and first- and last-mile connections determine whether the journey can be completed at all.

For transport policy, the relevant question is which barrier an intervention removes, and for whom. Delhi’s experience shows that reducing the fare changed behaviour substantially, while also revealing the conditions that determine how far that change can go.

The UK Walk to Work Trial

Community-wide programmes work with how people already travel by changing several conditions of the town at once. A narrower approach targets an individual directly, aiming to persuade a specific person to walk or cycle more. A large UK trial tested whether that narrower approach works on its own.

The Travel to Work trial recruited 654 employees across 87 workplaces in south-west England and south Wales and randomly assigned some workplaces to a ten-week intervention, delivered by trained workplace promoters, combining information provision, intention formation, identifying barriers and solutions, goal setting, self-monitoring, encouragement, social support, reviewing goals, and relapse prevention. Physical activity and travel mode were measured objectively, with accelerometers and GPS, at the start and again twelve months later (Audrey et al., 2019).

There was no evidence of an intervention effect on physical activity or modal shift at 12 months. The accompanying process evaluation identified distance to work, caring responsibilities, parking arrangements, and working conditions as important barriers to changing commuting behaviour (Audrey et al., 2019).

What the techniques could not do was create a workable journey out of a commute that was, for structural reasons, difficult to walk. Encouragement, goal setting, and self-monitoring addressed motivation. They left distance, caring responsibilities, and parking exactly as they were.

Stockholm’s Congestion Charge

A further question follows from the first three cases. If community-wide encouragement can shift routines at a town level, and individual persuasion alone often cannot shift a commute that is structurally difficult, what happens when a policy changes the cost of the old option directly, while also making the alternative better?

Stockholm’s congestion charge, introduced as a trial in 2006 and made permanent in 2007 after a referendum, applied broadly, with few exemptions, during charging hours. It was not introduced alone. Public transport services were expanded and new direct bus routes were added at the same time, and research on the trial found that this expansion was central to making the charge both workable and acceptable to the public (Kottenhoff and Brundell Freij, 2009). Traffic entering the charging zone fell substantially, and the reduction persisted and grew slightly over the following years once other factors were accounted for (Börjesson, Eliasson, Hugosson, and Brundell-Freij, 2012).

Stockholm produced a different outcome from the Walk to Work trial because it changed the underlying structure of the choice rather than relying on individual willpower, and because a genuinely improved alternative was available at the same time. The charge made driving more costly. The expanded bus network made the alternative better. The balance of costs had shifted, while a workable alternative was already there.

What the Cases Suggest

Four mechanisms run through these cases, each affecting a different point in the journey from policy to behaviour.

Community-wide change reaches people differently depending on what else is happening in their lives. The Sustainable Travel Towns programme produced its largest shifts among people whose routines were already changing for other reasons, and part of its effect took the form of a different destination or a trip that no longer happened, not only a different mode.

Usability has several separate dimensions, and a single intervention can move some of them without moving all of them. Delhi’s scheme, combining a fare waiver with added safety measures, measurably expanded when and how women could travel, while fleet reliability, the first mile, and the cost of Metro travel remained largely unchanged.

Motivation and feasibility are not the same thing. The Walk to Work trial gave people information, goals, and support, and produced no measurable change, because the techniques addressed willingness rather than the distance, caring responsibilities, and parking that made walking impractical for many participants.

Changing the structure of a choice works differently when the alternative is strengthened at the same time. Stockholm changed the cost of driving while strengthening the alternative, and the resulting shift proved durable in a way that persuasion alone, in the Walk to Work case, did not.

Prices, infrastructure, and rules still matter. They work by changing what a household’s routine has to pass through, and a policy reaches only as far as the conditions it actually changes.

Implications for Citizens, Institutions, and Policy

For citizens: A change in circumstances, a move, a new job, a change in who lives in the household, is a natural opening to try a different way of getting around. The habit that felt fixed before may not feel fixed anymore.

For institutions: Schools, employers, and hospitals set the shift times, start times, and locations that shape how people travel to them, even when transport is not their stated concern. The Walk to Work evidence suggests that workplace programmes aimed at persuasion alone will do little for staff whose commute is structurally difficult; addressing distance, parking, and scheduling matters as much as the encouragement itself.

For policymakers: Ask whether an intervention is changing the underlying structure of a choice or only trying to change what people want to do. Both can matter, and neither substitutes for the other. And when a policy is meant to expand what a group can do, ask which dimension of usability it actually moves, cost, safety, reliability, or connectivity, and which it leaves unchanged. Delhi’s scheme shows that solving two problems together, the fare and a safety concern, made a real difference for many women. It did not make every journey possible. Some women still face parts of the trip, like getting to the bus stop or affording the Metro, that the policy never touched.

Closing Note

The previous pieces in this series showed that sustainable choices take hold when they fit a problem the household already has. Transport adds one more layer: most of the time, people are not solving that problem freshly each day. They are running on habit, and habits are not moved by any single kind of intervention.

The barrier to sustainable transport is therefore not always resistance to greener modes. Sometimes it is an established routine that has not yet been disrupted. Sometimes it is a constraint, often unrelated to price, that makes the alternative unusable for a particular person. And sometimes willingness is already there, but the structure of the choice, the distance, the parking, the cost of the old option, has not changed enough to make a different routine possible.

The next installment will examine sustainable consumption, asking what the evidence shows about the gap between households’ stated intentions to consume less and the consumption patterns that actually unfold.

References

Audrey, S., Fisher, H., Cooper, A., Gaunt, D., Garfield, K., Metcalfe, C., Hollingworth, W., Gillison, F., Gabe-Walters, M., Rodgers, S., Davis, A.L., Insall, P., and Procter, S. (2019). Evaluation of an intervention to promote walking during the commute to work: a cluster randomised controlled trial. BMC Public Health, 19(1), 1-13. https://doi.org/10.1186/s12889-019-6791-4

Börjesson, M., Eliasson, J., Hugosson, M.B., and Brundell-Freij, K. (2012). The Stockholm congestion charges—5 years on. Effects, acceptability and lessons learnt. Transport Policy, 20, 1-12. https://doi.org/10.1016/j.tranpol.2011.11.001

Jamba, H., Devaraj, A., and Kanuri, C. (2025). Fare-Free Bus Travel Scheme for Women: Lessons from Delhi. WRI India Working Paper. https://wri-india.org/research/fare-free-bus-travel-scheme-women-lessons-delhi

Kottenhoff, K., and Brundell Freij, K. (2009). The role of public transport for feasibility and acceptability of congestion charging: The case of Stockholm. Transportation Research Part A: Policy and Practice, 43(3), 297-305. https://doi.org/10.1016/j.tra.2008.09.007

Ratra, V., and Kumar, R. (2026). Does free travel for women in public transportation make them more inclusive? An investigation of gendered subsidy initiative in India. Research in Transportation Economics, 116. https://doi.org/10.1016/j.retrec.2026.101713

Sil, A., Chowdhury, S., and Thoreau, R. (2024). How do you travel? A holistic evaluation of public transport journeys of women: A case study of Delhi, India. Journal of Public Transportation, 26. https://doi.org/10.1016/j.jpubtr.2024.100106

Sloman, L., Cairns, S., Newson, C., Anable, J., Pridmore, A., and Goodwin, P. (2010). The Effects of Smarter Choice Programmes in the Sustainable Travel Towns: Summary Report. Report to the Department for Transport. Transport for Quality of Life Ltd. https://webarchive.nationalarchives.gov.uk/ukgwa/20111005180138/http:/www.dft.gov.uk/publications/the-effects-of-smarter-choice-programmes-in-the-sustainable-travel-towns-summary-report

Verplanken, B., and Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy and Marketing, 25(1), 90-103.https://doi.org/10.1509/jppm.25.1.90

Verplanken, B., Walker, I., Davis, A., and Jurasek, M. (2008). Context change and travel mode choice: Combining the habit discontinuity and self-activation hypotheses. Journal of Environmental Psychology, 28(2), 121-127. https://doi.org/10.1016/j.jenvp.2007.10.005

As a result, incorporating sustainability into hospitality practices is vital for realizing responsible tourism growth and contributing to the United Nations Sustainable Development Goals (SDGs), specifically SDGs 6 (Clean Water and Sanitation), 7 (Affordable and Clean Energy), 8 (Decent Work and Economic Growth), 12 (Responsible Consumption and Production), and 13 (Climate Action) (UNEP, 2018; UN Tourism, 2021).

What is Green Hotel

Eco-hotels, or green hotels, are lodging facilities that implement eco-friendly and socially aware measures in their operations while ensuring excellent experiences for guests. Their main goal is to reduce adverse environmental effects, preserve natural resources, assist local communities, and guarantee sustainable economic success. In contrast to traditional hotels that mostly focus on operational efficiency for financial gain, green hotels embed sustainability into all facets of their business strategies from architectural design and energy usage to sourcing, waste disposal, and community involvement (UN Tourism, 2021).

The idea of eco-friendly hotels is directly connected to the tenets of sustainable tourism, which the United Nations World Tourism Organization (UN Tourism) describes as tourism that takes into account its existing and forthcoming economic, social, and environmental effects while meeting the requirements of tourists, the sector, the environment, and local communities (UN Tourism, 2021). In reality, this indicates that eco-friendly hotels aim to harmonize financial gain with environmental care and social accountability, making certain that hospitality services benefit both the community and the Earth.

Eco-friendly hotels adopt numerous sustainability measures. These typically involve enhancing energy efficiency via LED lighting, intelligent climate control systems, and renewable energy sources saving water through low-flow fixtures, rainwater collection, and wastewater treatment; reducing waste by cutting down single-use plastics, encouraging recycling, and composting organic waste; and obtaining food and services from local, ethical, and environmentally conscious suppliers. Numerous individuals also emphasize biodiversity preservation, eco-friendly cleaning supplies, sustainable gardening, and educational initiatives that motivate guests to engage in conservation activities (International Finance Corporation [IFC], 2017).

To promote transparency and credibility, numerous eco-friendly hotels willingly acquire globally acknowledged sustainability certifications. Initiatives like Green Key, Green Globe, Earth Check, and the certification criteria from the Global Sustainable Tourism Council (GSTC) evaluate hotels based on strict environmental, social, cultural, and management benchmarks. These credentials offer confidence.

Why Sustainability Matters in the Hospitality Industry

Sustainability has emerged as a crucial focus for the hospitality sector since hotels rely directly on natural assets, robust ecosystems, and flourishing local communities to function effectively. Hotel operations demand significant energy and water for heating, cooling, lighting, laundry, food preparation, and guest services. They also produce solid waste and wastewater as a result of their daily operations. With the continued growth of tourism worldwide, implementing sustainable practices is crucial to mitigate environmental effects and to guarantee the lasting viability of destinations and hospitality enterprises (United Nations Environment Programme [UNEP] & United Nations World Tourism Organization [UNWTO], 2005).

Environmental sustainability represents just one facet of accountable hospitality. The sector also holds significant economic and societal obligations. Hotels boost local economies by generating jobs, acquiring products and services from nearby vendors, and drawing in visitors who benefit neighboring businesses. Simultaneously, mismanaged tourism can exert stress on natural resources, cultural heritage, and local populations. Sustainable hospitality aims to enhance economic and social advantages while reducing negative environmental and cultural effects, aligning with the sustainable tourism principles advocated by UN Tourism.

From a business standpoint, sustainability has gained significant importance. Tourists are increasingly considering the environmental and social practices of lodging providers when deciding on their travel plans. Booking.com’s 2024 Sustainable Travel Report revealed that 83% of travelers value sustainable travel, and 75% intend to travel more sustainably in the upcoming year. This increasing consumer awareness drives hotels to implement trustworthy sustainability practices and clear sustainability certifications to accommodate evolving market demands.

Adopting sustainable methods can also enhance operational efficiency. Actions like energy-saving lighting, water conservation technologies, waste minimization programs, and sustainable purchasing assist hotels in utilizing resources more efficiently and can lower operating expenses in the long run. Moreover, globally acknowledged sustainability certification programs promote ongoing advancements in environmental management, resource efficiency, and social accountability, enhancing both corporate resilience and consumer trust.

Moreover, sustainability connects the hospitality sector with wider international development goals. Through the preservation of natural resources, minimizing waste, aiding local communities, encouraging responsible consumption, and participating in climate initiatives, green hotels contribute to multiple United Nations Sustainable Development Goals (SDGs), especially SDG 6 (Clean Water and Sanitation), SDG 7 (Affordable and Clean Energy), SDG 8 (Decent Work and Economic Growth), SDG 12 (Responsible Consumption and Production), and SDG 13 (Climate Action). With governments, businesses, and travelers placing greater emphasis on sustainability, incorporating responsible practices into hotel operations is becoming a vital element of lasting competitiveness and destination resilience.

Key Sustainable Practices in Green Hotels

Energy Efficiency and Renewable Energy

Energy conservation is part of the environmental criteria in the Global Sustainable Tourism Council (GSTC) Hotel Standard. The standard specifies that hotels must track energy usage by type, implement measures to reduce total energy consumption, and enhance the utilization of renewable energy sources.

To showcase implementation, the GSTC proposed performance indicators advise hotels to track overall energy usage and energy usage per guest night (or tourist night) for every energy source. Hotels are urged to track the percentage of renewable energy in their total energy supply and set objectives for minimizing energy use.

The GSTC also advises hotels to implement gear and operational methods that reduce energy usage and offer direction to employees and guests on decreasing unnecessary energy expenditure. These indicators aim to assist hotels in proving adherence to the GSTC Hotel Standard, while also permitting flexibility in choosing technologies and management strategies suited to their specific operational circumstances.

The Hotel Standard from the GSTC outlines the goals hotels should aim for, while the recommended performance indicators offer assistance in assessing adherence. The guidelines aim to create a shared international comprehension of sustainable tourism and serve as a benchmark for education, policy formulation, sustainability evaluation, and certification by GSTC endorsed certification organizations. GSTC does not directly certify hotels; instead, it accredits independent certification organizations that perform certification evaluations.

Water Conservation

Water conservation is incorporated as an environmental standard in the Global Sustainable Tourism Council (GSTC) Hotel Standard. The standard indicates that water risk must be evaluated, water usage should be quantified by category, and measures should be implemented to reduce total water use. It additionally mentions that obtaining water must be sustainable and should not negatively impact environmental flows. In regions recognized as having significant water risk, hotels are anticipated to establish and work towards context-specific water stewardship objectives.

The performance indicators recommended by the GSTC offer direction for assessing adherence to this criterion. They suggest that hotels evaluate and record water risk, track water usage per guest (tourist) night for every water source, set targets for decreasing water consumption, and track advancements toward those targets. The guidelines further suggest that hotels implement tools and procedures that reduce water usage and offer instructions to employees and visitors on how to use water responsibly.

The performance indicators additionally stipulate that hotels must source water from legitimate and sustainable origins that do not negatively impact environmental flows. They also suggest taking into account the combined effects of tourism on regional water supplies while overseeing water consumption. These performance metrics aim to offer direction for showcasing adherence to the GSTC Hotel Standard while permitting variation in how hotels reach the necessary results.

Alongside the Hotel Standard, the GSTC Certification Requirements for Performance in the Hotel/Accommodation Sector set forth essential performance criteria for hotels pursuing GSTC certification. For conserving water, these standards encompass practices like utilizing water-saving fixtures and appliances, effective water management in outdoor activities, and establishing linen reuse programs for guests who stay multiple nights. These actions are included in the certification criteria and aim to enhance water management in hotels that are certified.

Sustainable Food and Responsible Sourcing

Sustainable procurement is incorporated as a management standard in the Global Sustainable Tourism Council (GSTC) Hotel Standard. The guideline specifies that hotels need to adopt a purchasing policy that favors eco-friendly, locally suitable, fair trade, and locally manufactured products and services, whenever they are accessible. The policy is a component of the organization’s comprehensive sustainability management system and aims to influence purchasing choices throughout hotel operations.

The GSTC Industry Standard offers recommended performance metrics to assist in executing this criterion. These indicators suggest that hotels should track the percentage of goods and services acquired from local vendors, fair-trade sources, and other environmentally and socially responsible suppliers. They additionally suggest setting quantifiable goals and tracking advancements in enhancing sustainable buying practices.

The proposed performance indicators additionally advise prioritizing reusable, returnable, recycled, and recyclable products when suitable. Hotels are also urged to recognize procurement categories that could have notable environmental or social effects like items such as seafood, paper and wood products, textiles, and chemicals and set measurable objectives to enhance procurement practices in these sectors.

The GSTC Certification Requirements for Performance in the Hotel/Accommodation Sector offer further direction for hotels pursuing GSTC certification. These criteria facilitate the application of the sustainable purchasing standard by necessitating hotels to show documented procurement practices that align with the GSTC Hotel Standard and its related performance criteria.

Example - Six Senses Laamu, Maldives

An exemplary instance of sustainable hospitality is Six Senses Laamu. Sustainability is woven into the resort’s operations via efforts that emphasize water conservation, ethical waste management, sustainable sourcing, marine preservation, and community involvement. The resort has taken steps to minimize single-use plastics by introducing refillable water systems, offering plastic-free guest amenities, and engaging suppliers in initiatives that promote non-plastic packaging options. It additionally promotes sustainable local fishing and obtains various products from nearby communities when suitable.

A unique aspect of the resort is the Maldives Underwater Initiative (MUI), a joint program that includes Six Senses Laamu’s marine experts along with the Manta Trust, Blue Marine Foundation, and the Olive Ridley Project. MUI states that its goal is to enhance marine conservation via research, education, and community engagement. The project involves actions such as monitoring and restoring coral reefs, conserving sea grass, researching manta rays and sea turtles, and providing environmental education for visitors and local residents.

The resort has also created the Sea Hub of Environmental Learning in Laamu (SHELL), functioning as a center for marine science, conservation, and environmental education. SHELL backs research carried out by the Maldives Underwater Initiative and offers educational programs and engaging activities for guests, students, and local communities. These efforts showcase how a hotel can merge hospitality services with biodiversity preservation and environmental education via enduring collaborations and science-driven management.

Opportunities of Green Hotels

The shift to eco-friendly hotels offers chances for the hospitality sector to enhance environmental practices while fostering more sustainable tourism growth. Global frameworks established by the Global Sustainable Tourism Council (GSTC), UN Tourism, UNEP, and the International Finance Corporation (IFC) promote hotels to incorporate sustainability into their practices to improve resource efficiency, assist local communities, and aid in ongoing sustainable development.

Growing Demand for Sustainable Tourism

UN Tourism considers sustainability a key principle of tourism development and urges tourism enterprises to implement practices that reduce environmental effects while enhancing social and economic advantages for local communities. With sustainable tourism becoming more embedded in tourism policies and business strategies, hotels have a chance to integrate responsible environmental and social practices into their operations and support sustainable destination growth.

Cost Saving through Resource Efficiency

Enhancing resource efficiency offers hotels a chance to decrease energy, water, and other natural resource usage while boosting operational performance. The International Finance Corporation (IFC) advocates for resource-efficient methods in the tourism and hospitality industry as a facet of sustainable investment and environmental stewardship, acknowledging that the effective utilization of resources can yield enduring economic and ecological advantages.

Enhance Market Reputation and Recognition

Hotels that adopt globally acknowledged sustainability standards can showcase their dedication to responsible tourism via independent certification. The GSTC Hotel Standard acts as an internationally acknowledged guideline for sustainable hotel operations and underpins the certification process by accredited organizations. Certification offers independent confirmation that hotel operations are evaluated based on globally acknowledged sustainability standards, fostering increased trust among tourism stakeholders.

Supports Local Development

Sustainable hotels can boost local economies by emphasizing local jobs, sourcing products and services from nearby vendors, and aiding local businesses. UN Tourism and UNEP recognize that local sourcing, community involvement, and the retention of tourism benefits in destination communities are key elements of sustainable tourism that support inclusive economic growth.

Challenges of Green Hotels

Adopting sustainable methods in hotel operations necessitates ongoing dedication, efficient management, and persistent enhancement. The Global Sustainable Tourism Council (GSTC) recognizes sustainable management as a key necessity for hotels, highlighting the importance of incorporating environmental, social, cultural, and economic factors into operational planning and decision-making. Likewise, UNEP, UN Tourism, and the International Finance Corporation (IFC) acknowledge that realizing sustainability in the hospitality industry necessitates suitable investment, involvement of stakeholders, and systematic oversight of environmental and social effects.

High Initial Investment Cost

A key difficulty is the upfront funding needed to implement sustainable technologies and systems. Enhancements like energy-saving devices, renewable energy technologies, water-saving installations, and eco-friendly building renovations may necessitate significant capital investment. The IFC observes that making investment choices in sustainable hotels typically requires weighing initial expenses against future operational and financial advantages, thus making access to funding a crucial factor for numerous hotel operators.

Staff Capacity Building

The effective management systems and involvement of employees are crucial for the successful execution of sustainable practices. The GSTC Hotel Standard advises hotels to implement a formal sustainability management system, delegate responsibilities, offer suitable staff training, and consistently assess sustainability performance. Cultivating the knowledge, skills, and organizational capabilities necessary to fulfill these requirements can be difficult, especially for hotels that have restricted human and financial resources.

Integration of Sustainability in Daily Operations

Incorporating sustainability into daily hotel functions involves including environmental, social, cultural, and economic factors in planning, purchasing, resource management, and involvement with stakeholders. UNEP and UN Tourism highlight that sustainable tourism aims for a suitable equilibrium among environmental preservation, economic sustainability, and socio-cultural health. Consistently integrating these principles throughout all operational activities necessitates ongoing planning, execution, and assessment.

Monitoring and Continuous Improvement

Sustaining performance in sustainability necessitates methodical tracking and ongoing enhancement. The GSTC Hotel Standard advises hotels to track and assess metrics like energy usage, water consumption, waste production, and eco-friendly purchasing while consistently evaluating their sustainability goals and outcomes. Gathering trustworthy data, keeping records, and assessing progress are vital elements of proficient sustainability management and certification.

Policy Framework for Green Hotels

Governments set the legal standards that hotels need to adhere to, whereas international benchmarks like the Global Sustainable Tourism Council (GSTC) Hotel Standard offer a universally accepted framework for managing hotels sustainably. As stated by the GSTC, its criteria are designed to act as a foundation for sustainability certification, assist businesses in enhancing their sustainability efforts, and offer governments and various organizations a starting point for creating sustainable tourism guidelines.

Compliance with Legal Requirement

Hotels must adhere to all relevant national and local laws and regulations. The GSTC Hotel Standard recognizes legal compliance as an essential aspect of sustainable hotel management, addressing issues like environmental protection, land use, public health, workplace safety, labor, and other relevant laws. Sustainability certification supports these legal requirements but does not substitute them.

Sustainability Certification

The GSTC Hotel Standard acts as the basis for sustainability certification programs instead of being a legal obligation. Independent certification bodies evaluate hotels pursuing certification, rather than the GSTC itself. These certification organizations can certify hotels based on the GSTC Hotel Standard or another GSTC recognized Standard that has been independently confirmed to meet the GSTC criteria.

Role of Government and Tourism Authorities

The GSTC indicates that its standards can be utilized by governments, tourism agencies, and additional entities when formulating sustainable tourism policies and programs at national or regional levels. Instead of substituting local laws, the GSTC Standards offer a universal framework that can be tailored to fit specific environmental, social, cultural, and economic circumstances.

Hotels in Environmentally Sensitive Area

Hotels located in protected zones or other environmentally delicate areas must adhere to all relevant legal and regulatory standards concerning biodiversity preservation, land-use management, environmental safeguarding, and cultural heritage. The GSTC Hotel Standard mandates that hotels recognize relevant legal obligations and integrate them into their sustainability management systems to mitigate adverse environmental effects.

Recommendations for Hotels to Go Green

Hotels aiming to shift towards sustainable practices must embrace a holistic strategy that combines environmental, social, and economic sustainability within their management approaches. According to recommendations from the Global Sustainable Tourism Council (GSTC), UN Tourism, United Nations Environment Programme (UNEP), and the International Finance Corporation (IFC), the subsequent suggestions can facilitate the adoption of green hotel practices:

Create a Sustainability Management System

Hotels ought to create a recorded sustainability management system that incorporates environmental, social, cultural, and economic goals into their functioning. Establishing quantifiable sustainability goals, consistently tracking advancements, and perpetually evaluating performance can assist in securing enduring enhancements in sustainability management.

Enhance Energy Oversight

Hotels need to consistently track energy usage, set goals for energy savings, and implement energy-efficient operational strategies. Renewable energy sources should be integrated wherever possible to decrease greenhouse gas emissions and enhance long-term resource efficiency.

Water Management Practices

To manage water effectively, hotels must track water usage, evaluate water-related risks, and adopt conservation strategies. Hotels must also guarantee that water is obtained sustainably and set water stewardship objectives, especially in areas facing water shortages.

Minimize Waste Production

Hotels need to establish a thorough waste management strategy that focuses on waste reduction, reuse, recycling, and proper disposal. Tracking waste production and cutting down on food waste can greatly lessen environmental consequences while enhancing operational effectiveness.

Embrace Eco-Friendly Procurement Methods

Hotels ought to establish sustainable procurement policies that focus on locally sourced, eco-friendly, and fair-trade products and services when accessible. Ethical buying practices can lessen environmental effects while promoting local economies and sustainable supply chains.

Assist Local Communities

Hotels should generate job opportunities for nearby residents, procure products and services from local enterprises whenever feasible, and honor local customs and traditions. Fostering connections with local communities’ aids in making sure that tourism supports inclusive and sustainable growth.

Train Employees and Raise Guest Awareness

Employees should receive consistent sustainability training to promote responsible operational practices. Hotels ought to inform guests about sustainability efforts and motivate them to engage in conserving energy, water, and other resources while they visit.

Conclusion

Eco-friendly hotels show how the hospitality sector can minimize its ecological footprint while delivering excellent guest experiences via responsible and sustainable methods. Hotels can enhance responsible tourism and ensure long-term business resilience by boosting energy and water efficiency, implementing sustainable sourcing, reducing waste, aiding local communities, and adhering to globally recognized sustainability standards. Despite obstacles like substantial upfront costs, employee training, and ongoing oversight, the increasing interest in sustainable tourism, favorable policy initiatives, and international certification benchmarks offer considerable prospects for the industry. With sustainability becoming essential in hospitality, eco-friendly hotels will take on a growing significance in promoting responsible tourism and aiding wider sustainable development objectives.

References

Booking.com. (2024). Sustainable travel report 2024. https://www.booking.com/sustainable-travel/

Global Sustainable Tourism Council. (2025). GSTC industry criteria with performance indicators for hotels and accommodations.

https://www.gstc.org/

Global Sustainable Tourism Council. (2026a). GSTC certification requirements for performance in the hotel/accommodation sector. https://www.gstc.org/wp-content/uploads/GSTC-Certification-Requirements-for-Performance-in-the-Hotel_Accommodation-Sector-April-2026.pdf

Global Sustainable Tourism Council. (2026b). GSTC hotel standard. https://www.gstc.org/gstc-criteria/gstc-hotel-standard/

Global Sustainable Tourism Council. (2026c). GSTC industry standard for hotels: Suggested performance indicators. https://www.gstc.org/gstc-criteria/gstc-industry-criteria-for-hotels/

Global Sustainable Tourism Council. (2026d). GSTC-recognized standards for hotels. https://www.gstc.org/gstc-criteria/gstc-recognized-standards-for-hotels/

Global Sustainable Tourism Council. (2026e). For hotels & accommodations. https://www.gstc.org/for-hotels-accommodations/

International Finance Corporation. (2017). Environmental, health, and safety guidelines for tourism and hospitality development. World Bank Group.

https://www.ifc.org/

International Finance Corporation. (2020). The business case for sustainable hotels. World Bank Group. https://www.ifc.org/en/pressroom/2020/the-business-case-for-sustainable-hotels-new-report-launched-by

Maldives Underwater Initiative. (2026). About the Maldives Underwater Initiative. https://www.maldivesunderwaterinitiative.com/about-mui

Six Senses. (2023). Six Senses Laamu celebrates the inauguration of the Sea Hub of Environmental Learning in Laamu (SHELL). https://www.sixsenses.com/en/corporate/media-center/press-releases/2023/six-senses-laamu-sustainability-hub/

Six Senses. (2026). Sustainability in Laamu. https://www.sixsenses.com/en/hotels-resorts/asia-the-pacific/maldives/laamu/sustainability/

United Nations Environment Programme. (2018). Tourism. https://www.unep.org/explore-topics/resource-efficiency/what-we-do/responsible-industry/tourism

United Nations Environment Programme, & United Nations World Tourism Organization. (2005). Making tourism more sustainable: A guide for policy makers. https://www.unep.org/resources/report/making-tourism-more-sustainable-guide-policy-makers

UN Tourism. (2021). What is sustainable tourism? https://en.unwto-ap.org/faq/10712/

This chain, running from the mine to the scrapyard, falls under what accountants call “Scope 3”: every indirect emission across a company’s value chain that isn’t part of its own direct operations. It’s also the biggest piece of the puzzle, usually making up 70–90% of a company’s total carbon footprint. Yet because it cuts across multiple layers of suppliers and countries, it has escaped serious regulatory or market scrutiny. The semiconductor and AI hardware industry is an extreme case: Scope 3 dominates its footprint, its supply chain is scattered across the globe, and its governance structures are still underdeveloped. Most past discussions of Big Tech’s ESG performance have stopped at the company gate, often failing to pull mineral extraction, wafer fabrication, product use, and end-of-life disposal into one picture. That gap is what this article tries to fill.

Why Scope 3 Exposes a Company’s Structural Choices

To understand the carbon footprint of AI supply chains, we first need to understand how the greenhouse gas accounting system itself draws its boundaries. Under the GHG Protocol, Scope 1 covers emissions from sources a company directly controls (like the process gases released during wafer etching). Scope 2 covers indirect emissions from purchased electricity and steam. Scope 3 is broken into fifteen categories, spanning upstream activities like “purchased goods and services” and “capital goods” and “upstream transportation” down to downstream activities like “use of sold products” and “end-of-life treatment”. One key feature of this system: how vertically integrated a company is directly determines which scope its emissions will mostly fall under.

Nowhere is this clearer than in semiconductors. TSMC, as an integrated foundry, handles its own etching, cleaning, and lithography from start to finish, so its Scope 1 emissions (direct process gases) plus Scope 2 emissions (purchased electricity for its fabs) together account for over 60% of its total footprint. Nvidia, by contrast, is a classic “fabless” design company: manufacturing is outsourced entirely to TSMC and packaging and testing to ASE. Nvidia’s own Scope 1 and 2 emissions make up less than 4% of its total, while Scope 3 accounts for 96.6%, and within that, “purchased goods and services” alone makes up 87% of its Scope 3 footprint. The difference between these two business models shapes where each company’s emissions report puts its frame. The less vertically integrated a company is, the more easily its emissions responsibility gets pushed through outsourcing into Scope 3 categories that sit outside the company’s own financial reporting boundary, while its Scope 1 and 2 numbers look comparatively impressive. The accounting boundaries are simply drawn this way; it isn’t deliberate manipulation. But the practical effect is that “fabless” companies end up with an easier operational emissions story to tell.

This structural pattern overlaps with a broader trend: as multinational companies expand production capacity globally through foreign direct investment (FDI), they also shift where carbon emissions physically happen. A research team at Tianjin University used input-output modeling and structural decomposition analysis to track the embedded carbon emissions in multinational supply chains over a long time series, finding that between 2005 and 2016, changes in these embedded emissions were driven mainly by three factors: scale effects (output growth), technology effects (process improvements), and intensity effects (changes in emissions per unit of output). The relative weight of each factor shifted across different periods: between 2005 and 2008, scale effects accounted for about 20% of emissions growth; after 2011, the trend reversed, with intensity improvements becoming the dominant factor.

This same framework helps explain what’s happening in AI hardware supply chains. The wave of new fabs that TSMC, Samsung, and Intel have built in the U.S., Japan, and Europe recently is, at its core, a cross-border transfer of both production capacity and carbon emissions via FDI. The embedded emissions of a new fab are counted, on paper, as operational emissions of the country where it’s built. But the capital, technology, and order volume driving that construction originate from a different jurisdiction, wherever the chip designers and brand owners are based. This geographic mismatch between who emits and who benefits is one of the structural reasons global supply chain carbon governance still struggles to pin down accountability.

Upstream

Semiconductors and AI hardware depend on a fixed set of critical minerals: silicon, copper, cobalt, nickel, lithium, tungsten, and the seventeen metals collectively known as “rare earth elements.” These minerals are distributed unevenly around the world, which is why upstream governance is the weakest link in the AI supply chain.

The Democratic Republic of Congo supplies roughly 70% of the world’s cobalt, and somewhere between 10% and 30% of that comes from artisanal and small-scale mining operations with little to no basic safety protection. Cave-ins, suffocation, and child labor have been documented by international human rights organizations. Years ago, the OECD already published its Due Diligence Guidance for Responsible Supply Chains of Minerals from Conflict-Affected and High-Risk Areas, aiming to give downstream buyers a methodology for identifying, assessing, and mitigating human rights and governance risks in mineral supply chains. The International Council on Mining and Metals (ICMM) and the Initiative for Responsible Mining Assurance (IRMA) have each rolled out their own industry sustainability standards and third-party certification systems. But most of these frameworks rely on voluntary participation, which gives them little real traction over a scattered, informal artisanal mining sector. Cobalt ore travels from Congo to China for chemical processing, passing through multiple traders along the way, and information about its original source gets murkier with every handoff. This remains one of the hardest traceability problems in battery and semiconductor supply chain due diligence.

Rare earths present a different governance picture altogether. China accounts for roughly 70% of global rare earth mining and 90% of refining and processing, which gives it de facto pricing power and technical control over this entire segment of the supply chain, while also leaving it to shoulder the tailings management and water pollution burdens that come with extraction and ore processing. Meanwhile, cobalt mining regions in Congo face a double bind of forced relocations and contaminated community water sources. Lithium mining in Zimbabwe and graphite mining in Madagascar and Mozambique are also seeing a fresh wave of development driven by AI hardware demand. Some African research institutions have called this new round of resource competition a defining moment of the AI era: it’s a window for resource-rich countries to gain more bargaining power and push for local value-added processing, but it also carries the risk of repeating the “resource curse” pattern of past decades: raw materials sold cheap, the value-added processing happening elsewhere, and the environmental and social costs left to pile up at home.

From ore to chip, this chain carries more than carbon emissions: it also carries a toxicology profile that’s long been overlooked. A chemical analysis of the elemental composition of Nvidia’s A100 GPU found that elements classified as hazardous made up as much as 93% of the card, including more than a dozen substances officially listed as toxic: arsenic, mercury, lead, cadmium, chromium, cobalt, and beryllium, among others. In other words, every GPU used to train a large model has carried a share of hazardous materials since the moment its ore was smelted, and this piece of the environmental cost, incurred at the earliest stage of the product’s life cycle, almost never shows up on any AI company’s carbon ledger. The mineral volume consumed by a single training run is negligible against global annual mineral output. But the scale worth scrutinizing is the cumulative effect of thousands of training runs and hardware refresh cycles stacking on top of each other.

This governance mismatch is just as visible at the international level. In late 2025, the U.S. State Department led the formation of a nine-country strategic alliance called “Pax Silica,” bringing together Japan, South Korea, Singapore, the Netherlands, the U.K., Israel, the UAE, and Australia, with the stated goal of securing AI, semiconductor, and critical mineral supply chains and reducing dependence on China. The alliance didn’t include a single “Global South” country at launch, even though these are the countries where most critical minerals are actually extracted; India was added later. That structural exclusion sends a blunt signal: in the international division of labor around AI hardware supply chains, resource-rich countries are still cast mainly as raw material suppliers rather than equal partners in governance. In early 2026, the U.S. went a step further, hosting a ministerial-level meeting on critical minerals and signing partnership agreements with Argentina, the Cook Islands, Ecuador, Guinea, and Morocco. So far, this “de-risking” approach to mineral diplomacy hasn’t shown any clear connection to improvements in environmental or human rights governance standards. In a recent working paper, the OECD described this shift as mineral supply chains moving from being “market-driven” to “policy-driven”: governments are now shaping supply chains directly through subsidies, tax credits, and equity investment. But faster industrial policy doesn’t necessarily mean governance standards are keeping pace. Most countries already have forecasting scenarios for energy consumption but still lack equivalent models for material consumption, meaning policymakers’ read on demand-side pressures still lacks systematic scientific grounding.

Midstream

While mineral extraction remains the most obscured link in the supply chain, wafer fabrication stands out as the most energy-intensive. It is also the hardest stage to decarbonize through efficiency gains alone. Modern advanced-node fabs consume power on a scale rivaling mid-sized chemical plants. This energy appetite cannot be engineered away; given the physical limits of advanced manufacturing, energy consumption per unit of chip area will climb rather than decline.

Beyond raw power consumption, a technical nuance often escapes outside observers: the true climate impact of a fab’s scope 1 emissions comes from the Global Warming Potential (GWP) of the specific process gases involved, not the sheer volume of gas released. Fluorinated gases vital to etching and cleaning, such as per-fluorocarbons (PFCs), nitrogen trifluoride (NF₃), and sulphur hexafluoride (SF₆), carry GWPs thousands or even tens of thousands of times higher than carbon dioxide. As a result, even a minuscule leak translates into a large carbon footprint when converted to CO₂-equivalent terms.

This impact explains why Scope 1 emissions remain a priority for industry reduction efforts, even though they comprise only 9% of TSMC’s total footprint. To tackle this leverage point, the foundry sector currently relies on three technical avenues: installing point-of-use local scrubbers, transitioning to carbon-neutral natural gas, and pioneering alternative low-GWP process gases.

TSMC has the most complete data disclosure in this entire supply chain. In its most recent fiscal year, the company’s total energy consumption came to 27,456 gigawatt-hours, with purchased electricity making up 93% of that. It carried out 1,177 energy-saving measures over the year, saving 810 gigawatt-hours of electricity, equivalent to cutting 400,000 tonnes of carbon emissions. TSMC’s broader strategy is representative of where the industry is headed: starting in 2025, carbon reduction performance officially became a factor in supplier selection. Suppliers tied to key emission sources now have to sign greenhouse gas reduction agreements, committing to 85% renewable energy for Taiwan-based capacity and 100% for overseas capacity by 2030 and to adopting science-based reduction targets by 2035. More than fifty suppliers have already signed on, covering nearly 90% of TSMC’s supply chain emissions. The company has also launched a supply chain decarbonization subsidy program, helping first-tier raw material suppliers upgrade to more energy-efficient equipment, expected to cut 450,000 tonnes of emissions, roughly the annual carbon sequestration of 45,000 hectares of forest. For TSMC’s upstream material and equipment suppliers, ESG performance has stopped being a nice-to-have and has become a condition for staying on the order list at all.

Even so, TSMC still can’t escape the scale pressure of its own expanding capacity. In 2026, for the first time, its emissions in Taiwan will exceed the statutory threshold and trigger a carbon fee. Samsung Electronics tells a more contradictory story. Its consumer electronics division (DX) has already hit a 93.4% renewable energy transition rate, while its semiconductor division (DS) sits at just 24.8%, a gap relative to the company’s overall 2030 target of 100% renewables. DX has committed to net zero by 2030; DS’s net-zero target isn’t until 2050, a twenty-year gap between the two. There’s a technical reason behind this “two-track” setup: process gas emissions in semiconductor manufacturing are more complex to manage than those in consumer electronics. But the practical result is that Samsung’s fastest-growing, highest-emitting division also happens to be the one operating on the most relaxed governance timeline. Samsung’s operational emissions (Scope 1 and 2) rose 7.83% year-over-year in its most recent reporting period, driven mainly by expanding semiconductor capacity and rising AI-related manufacturing demand. Its total carbon footprint, including Scope 3, tops 125 million tonnes. More than 80% of that Scope 3 figure comes from downstream “product use”, and the company has yet to set an absolute Scope 3 reduction target. Compared with TSMC and Intel, Samsung’s renewable energy procurement has long leaned on lower-quality renewable energy certificates rather than power purchase agreements, which deliver more substantive emissions reductions. That’s a big part of why outside analysts have questioned the “internal credibility” of Samsung’s sustainability reporting.

Infineon and UMC’s cooperation memorandum, signed in March 2026, points toward a different approach: rather than each company going it alone, why not get shared suppliers to align on a single, unified set of science-based reduction targets? Joint workshops and shared best practices help bring down compliance costs for small and mid-sized suppliers. UMC’s “Supply Chain Carbon Inventory Coaching Program,” running since 2022, has already reached more than 400 suppliers. Infineon, for its part, has committed to cutting Scope 1 and 2 emissions 72.5% below its 2019 baseline by 2030 and has set a Scope 3 target too: getting 72.5% of suppliers to adopt science-based reduction targets covering purchased goods and services, capital equipment, and transportation and distribution by 2029. GlobalFoundries also accelerated its commitments in 2025, raising its reduction target from 25% to 42% against a 2021 baseline, and is still pursuing external verification even as global semiconductor capacity keeps expanding. This model, where a major player takes the lead and industry associations standardize the process, is a response to a deeper structural problem: roughly 70% of suppliers worldwide are small or mid-sized businesses that lack both the specialized staff and the data infrastructure to put together a compliant carbon footprint accounting. Without funding and technical support from bigger companies, “supply chain decarbonization” risks staying a paper exercise. It’s telling that it took the Semiconductor Climate Consortium’s Scope 3 working group a year and a half just to draft a unified industry guideline for Category 11 accounting (“use of sold products”), a sign of how complicated this work is.

Equipment makers matter here too. ASML has set an explicit target of net zero across its entire value chain by 2040 and, by 2030, wants 75% of its top suppliers, ranked by emissions, to have made formal reduction commitments. Texas Instruments has pledged 100% renewable energy for its 12-inch wafer fabrication operations by 2025. TSMC pioneered a “joint renewable energy procurement” model in 2023: a twenty-year joint purchasing agreement covering 20 billion kWh that spreads the “green premium” across a wider slice of the value chain. It’s seen as a replicable way to solve the problem that individual procurement is simply too expensive for smaller suppliers to manage on their own. Once purchase orders start coming with carbon footprint requirements attached, ESG metrics effectively become part of the language of market access for equipment sales. A lithography tool or an etching machine that can’t produce competitive numbers on energy use and resource efficiency starts off at a real disadvantage in customer evaluations.

Downstream

The growth curve of data center electricity consumption is the easiest part of this supply chain to quantify, and the one that gets the most public attention. Global data center electricity use hit around 415 terawatt-hours in the most recent year, about 1.5% of global electricity consumption, growing at an average of 12% a year over the past five years. That figure is projected to double to 945 terawatt-hours by 2030, pushing its share up to 3%. But the more important shift is happening inside that number: electricity consumption at data centers dedicated to AI training and inference jumped 50% year-over-year in the most recent period, far outpacing the 17% growth rate for data centers overall, and is expected to triple again by 2030. Behind all of this sits more than $400 billion a year in combined capital spending from five major tech companies: Amazon, Google, Meta, Microsoft, and Equinox. That figure already exceeds total global investment in oil and gas production, and it’s expected to grow another 75%. Over the same period, natural gas and coal combined will still meet more than 40% of data centers’ new electricity demand through 2030. This means the AI compute boom is, in some regions, extending the grid’s reliance on fossil fuels rather than accelerating renewable buildout.

There’s also a methodological wrinkle in data center carbon accounting that the public rarely notices but that has an outsized effect on companies’ reported numbers. The GHG Protocol’s Scope 2 Guidance requires companies to disclose two separate sets of figures: a “market-based” calculation and a “location-based” one. The market-based method lets companies offset the carbon intensity of their actual electricity use with purchased renewable energy certificates or power purchase agreements (PPAs) and claim “carbon neutrality” on that basis. The location-based method, by contrast, uses the actual average carbon intensity of the grid where the data center sits, reflecting the real load the grid is carrying. The gap between these two figures is exactly the part that tends to get cherry-picked in corporate marketing materials. On a location-based basis, Google’s Scope 2 emissions have climbed from 5.8 million tonnes to over 11.2 million tonnes; Microsoft’s went from 4.3 million tonnes to nearly 10 million tonnes over the same period. These numbers rarely make it into the headlines of corporate press releases, which typically lead with the far more flattering market-based figures instead. This whole “market-based vs. location-based” debate is really just one piece of a bigger argument about the “additionally” of renewable energy purchases: buying green certificates from an already-built wind farm and financing the construction of a brand-new one have different effects on actual grid emissions, but under market-based accounting, the two often end up looking almost identical on paper.

At the company level, this tension between “clean operations and uncontrolled supply chain” shows up in data centers. Google’s data center electricity consumption rose 27% year-over-year in the most recent period, yet efficiency gains actually pushed data center emissions down 12%, the first time it’s managed to decouple energy use from emissions at that level. And yet the company’s overall Scope 3 emissions grew 22% that same year, pushing its total carbon footprint 51% above its 2019 baseline. Water use tells a similarly overlooked story: Google consumed about 3 billion liters of water across its data centers and offices in the most recent year, up 28% year-over-year, with more than a quarter of that drawn from water-stressed regions. Data centers alone accounted for roughly 27 billion liters of drinking-quality water. Over the same period, Google raised its water replenishment rate from 18% to 64%. That’s real progress, but it was achieved against consumption rising 28% over the same period, meaning replenishment is still running behind actual use. Estimates put the global AI sector’s annual water footprint somewhere between 312.5 billion and 764.6 billion liters, a scale large enough to keep fueling debate over whether it makes sense to site data centers in drought-prone regions.

Nvidia’s supply chain paints an even starker picture of geographic imbalance. Among its East Asian manufacturing suppliers, only 24.09% use renewable energy, compared with 80.55% among its North American suppliers: the same company and the same carbon neutrality pledge, but two different realities on the ground. AMD, Qualcomm, Nvidia, and Broadcom now see supply chain emissions account for more than 80% of their total footprints, and Nvidia in particular has yet to make any concrete commitment to renewable energy investment among its suppliers. South Korea’s government has approved a new 1-gigawatt LNG cogeneration plant for SK Hynix, with another 3 gigawatts of similar capacity planned for Samsung. Taiwan Power Company has also cited rising electricity demand from the semiconductor and AI industries as one justification for expanding its LNG projects and grid infrastructure. This means some of the new power feeding AI chip production will still come directly from fossil fuels.

Apple’s case offers a useful window into what happens when ESG narratives run into legal scrutiny. The company announced its global greenhouse gas emissions had fallen by more than 60% since 2015, emphasizing that the figure excluded carbon offsets. But not long after that announcement, a German court ruled that Apple could no longer market the Apple Watch as “carbon neutral.” The environmental group Deutsche Umwelthilfe had accused Apple of greenwashing, and the court found that Apple’s explanation of its carbon offset projects failed to give consumers a clear understanding of how effective those offsets actually were. Apple has publicly stated that its core assembly suppliers, Wistron and Pegatron, have significantly increased their renewable energy usage. But by Apple’s own reported figures, Pegatron’s renewable energy share stood at just 47% as of the end of 2023, a gap from the “nearly there” impression Apple’s broader narrative tends to convey. Apple also disclosed that its semiconductor and display suppliers cut a combined 8.4 million tonnes of emissions in the most recent year and that twenty-six direct semiconductor suppliers have committed to eliminating at least 90% of their fluorinated greenhouse gas emissions by 2030, one of the rare disclosures that puts the upstream process gas problem on the table. It also suggests companies are being noticeably selective about which categories of Scope 3 emissions they choose to be transparent about.

E-Waste Governance

AI hardware’s life cycle doesn’t end at the data center rack; it eventually has to be disposed of. And this is where international governance across the entire supply chain is weakest.

Global e-waste generation has already reached 62 million tonnes a year and is projected to hit 74 million tonnes by 2030, growing five times faster than recycling capacity can keep up. Worldwide, only about 17% to 22% of e-waste is properly collected and recycled; most of the rest ends up in landfills or gets shipped to developing countries with weaker environmental and labor protections, where it’s often broken down in informal workshops through open burning and acid baths to extract precious metals, releasing as many as a thousand different chemical substances and causing long-term contamination of local soil, water, and air. Cross-border e-waste movement is, in theory, governed by the Basel Convention and its Basel Ban Amendment, which prohibits exporting hazardous waste from OECD to non-OECD countries. But enforcement has long been undermined by uneven compliance capacity among member states and the sheer fragmentation of informal recycling networks, and the actual volume of e-waste flowing into developing countries far exceeds what the Convention can realistically track. In markets with relatively mature formal recycling systems, third-party certifications like R2v3 and e-Stewards provide chain-of-custody and data-destruction standards for IT asset disposal. But these certifications currently cover mainly the disposal processes companies directly control.

AI hardware is also being retired at a noticeably faster pace than traditional IT equipment. A recalculation based on TSMC’s actual CoWoS advanced packaging capacity found that some earlier forecasting models, which had projected e-waste from retired AI servers reaching hundreds of millions of kilograms annually and growing out of control, had overestimated the figure by seven to fourteen times. Based on a five-year service life, e-waste from AI servers is indeed growing at a notable clip, just not nearly as dramatically as earlier projections suggested. What’s actually worth watching is how much faster AI chips themselves cycle out compared with traditional servers: recently released “leading-edge AI chips” have an average useful life of just 3.9 years from launch to being phased out of frontier model training, and some companies have compressed their GPU cluster refresh cycles down to 18–36 months. Each new architecture generation delivers such a large performance jump that holding onto older hardware simply stops making economic sense.

Not all of this e-waste is a pure liability, though. Roughly one-fifth of global e-waste consists of recoverable metals, including millions of tons of cobalt, aluminum, and other critical materials: an estimated $60 billion worth of raw material value. In the U.S. alone, discarded electronics lose an estimated $4 billion worth of copper, palladium, gold, and other precious metals every year. That’s the commercial logic behind Apple and Microsoft’s push into “circular data center” programs. Microsoft has achieved a 90.9% reuse and recycling rate for server components, with more than 3.2 million components reused through internal or external channels. Apple’s disassembly robot “Daisy” can take apart thirty-six iPhone models, and the company now uses 100% recycled cobalt in its own battery designs and 100% recycled rare earth elements in its magnets. But the scope of these programs stops at the edge of a company’s own balance sheet. Once a GPU or server enters the longer, more fragmented world of secondhand refurbishment and teardown-for-parts recycling, almost no AI company folds that “downstream of downstream” environmental impact into its sustainability reporting. This accountability gap, layered on top of the Basel Convention’s structural enforcement weaknesses, makes the tail end of this supply chain the least regulated and most easily overlooked link in the entire chain.

What’s Actually Hidden in ESG Reports

Put the upstream, midstream, and downstream data side by side, and a pattern emerges that’s almost statistically predictable: companies disclose the most detail and show the most progress exactly where things are easiest to control and turn into PR material: office electricity use, data center PUE, and the recycled-material share of a flagship product. The areas where disclosure becomes vague and progress stagnates are precisely the ones that matter most: upstream mineral extraction, midstream process gas emissions, and genuine end-of-life disposal downstream.

This pattern holds up beyond casual industry observation, in large-scale empirical research. Accounting researchers tested this directly using a sample of over 35,000 publicly listed companies worldwide, covering nearly 200,000 customer-supplier-year observations, examining a signaling-theory hypothesis: do companies selectively disclose their environmentally strong suppliers while quietly hiding the polluting ones, effectively engaging in “supply chain greenwashing”? The findings were unambiguous. As a supplier’s environmental score moves from the bottom of the distribution toward the top, the odds that a downstream customer voluntarily discloses that relationship rise sharply. But this selective disclosure shows no statistical link to any real improvement in a company’s overall supply chain environmental performance. The more damning evidence comes from the opposite direction: companies’ newly added suppliers actually score lower on average for environmental performance, and even among the companies that disclose the most selectively, their new suppliers perform no better environmentally than their peers’. That directly undercuts the “effective communication” theory: the idea that companies are screening for greener suppliers and signaling real progress to the market. This behavior didn’t taper off after the Paris Agreement and the broader rise in ESG scrutiny; if anything, it grew more pronounced. The more external attention a company draws, the harder it works to hide the bad parts, which points to opportunistic greenwashing rather than genuine value communication. Most of this selective disclosure shows up in the business-description sections companies write voluntarily and in standalone ESG reports; polluting suppliers, by contrast, tend to appear only in the “risk factors” section that regulators require. Companies only mention them when legal or reputational exposure leaves no other choice.

Apply this framework to the AI hardware industry, and a lot of details start making sense. Companies are happy to devote pages to a supplier’s clean-energy installed capacity, or the recycled-material percentage of a “carbon-neutral” product. What they’re reluctant to show is the twenty-year gap between net-zero timelines for semiconductor divisions, the yawning difference in renewable electricity use between East Asian and North American suppliers, or the gap between market-based and location-based Scope 2 numbers. There’s also a structural asymmetry in how the narrative gets told, and it’s become standard industry practice: companies avoid discussing the downstream emissions that AI itself enables, like the indirect carbon footprint from AI being used to accelerate oil exploration or optimize logistics, while heavily promoting “emissions reduction potential” calculated from hypothetical scenarios. One tech company calculated that five of its own products, a smart thermostat and mapping and navigation tools among them, helped external users avoid roughly 26 million tonnes of emissions in a single year, equivalent to the annual electricity emissions of 3.5 million American households. That figure is close to the company’s own total annual emissions, yet there’s no clear path for independent third-party verification of it. This pattern of hiding the costs while amplifying the benefits is probably the most worrying feature of the current AI-and-ESG discourse. Capital markets respond to this kind of narrative in an uneven way, too: when a company discloses green suppliers, retail investors tend to push the stock price up, but that premium fades within two years and sometimes reverses entirely. Professional analysts’ earnings forecasts, by contrast, are essentially unaffected by this kind of disclosure from the start, suggesting that sophisticated market participants are far better at seeing through selective disclosure than ordinary investors are.

Why Capital Markets Have Made Carbon Emissions ESG’s “Hard Currency”

The recent acceleration in this catch-up effort, happening mostly over the past year or two, owes a lot to a shift in how capital markets think about the issue, and there’s a clear pricing logic underneath it. Carbon emissions became the first ESG dimension to get properly “monetized” mainly because they have three qualities that the S (social) and G (governance) sides of ESG generally lack: they can be converted, via global warming potential, into a single universal unit, CO₂-equivalent; they can be priced, traded, and plugged directly into financial models; and they connect directly to a company’s transition risk and physical risk, giving them clear financial materiality. That’s essentially why the International Sustainability Standards Board (ISSB) carved climate disclosure out as its own standard, IFRS S2, separate from IFRS S1, which covers sustainability topics more broadly. In mainstream ESG rating systems, climate and carbon emissions typically make up 30% to 50% of the total score, the largest single component. Vague or incomplete carbon data drags a rating down directly, which pushes up financing costs and drags down valuations. Institutions like ICBC, HSBC, and JPMorgan have already folded ESG performance into their credit approval or risk pricing models, companies with weak ESG scores can see loan rates bumped up by half a point to two full percentage points. For fabs and data centers with capital spending routinely running into the tens of billions, that’s not a difference anyone can afford to ignore.

For the semiconductor industry specifically, this capital market pressure is turning into a real fight over the “green premium.” Buying clean power and upgrading to more energy-efficient equipment inevitably pushes costs up in the short term as supply chains go green. TSMC’s twenty-year, 20-billion-kWh joint renewable procurement model is, at its core, a piece of financial engineering designed to spread that green premium out, taking a cost that no single small supplier could absorb on its own and distributing it across the whole supply chain. The carbon removal market shows a similar split. Technology-based removal methods, like direct air capture, typically run $170 to $500 per ton, seven to twenty times the cost of nature-based solutions. Big tech companies like Microsoft have already locked in forward purchase agreements covering more than 40 million tonnes, accounting for 80% of the entire market, even though spot market supply is only about 1.4% of that forward-contracted volume. Large, well-funded tech companies are essentially buying up the best carbon removal capacity years in advance, which risks pushing smaller suppliers and supply chain partners in developing countries even further to the margins. That also helps explain why chipmakers and equipment manufacturers are increasingly building ESG metrics directly into their product marketing: once buyers start factoring carbon footprint data into supplier selection, ESG performance stops being a nice PR add-on and becomes a precondition for actually winning the order. This kind of hard constraint, driven jointly by capital market pressure and supply chain pressure, is one of the few mechanisms in the entire AI supply chain carbon governance story that’s self-enforcing.

Global Governance Landscape

Facing a supply chain that spans mines, fabs, data centers, and teardown workshops, major economies and international bodies are responding at different speeds, and that divergence has become more visible.

The EU’s legislative framework used to be the most comprehensive and the most binding, but it’s undergone a significant political retreat over the past year. The Corporate Sustainability Reporting Directive originally required covered companies to extend disclosure obligations down through their supply chains, but the “Omnibus I” simplification package, moving through 2025 and 2026, sharply narrowed which companies the rules even apply to, by some estimates cutting the covered scope by roughly 80% compared with the original proposal. The Corporate Sustainability Due Diligence Directive saw an even more thorough rollback: the amended version finally approved by the EU Council in February 2026 removed the originally unified civil liability provision, narrowed due diligence obligations down to first-tier direct suppliers only, and pushed the overall implementation timeline back to 2029. Under the banner of “simplification,” this amounts to a systematic weakening of supply chain human rights and environmental due diligence. Multiple multinational companies and human rights organizations jointly called for keeping the original strength of the rules but couldn’t reverse the direction of the political process. By contrast, the Carbon Border Adjustment Mechanism (CBAM) and the Green Claims Directive have held up more intact. CBAM’s coverage is expanding from its original six categories (steel, cement, aluminum, fertilizer, electricity, and hydrogen) into downstream derivative products, and from 2027 it will also start covering indirect Scope 3 emissions. The Green Claims Directive, which took effect in March 2026, explicitly bans companies from using terms like “carbon neutral” or “low-carbon” without third-party verification, with penalties for violations reaching up to 4% of annual turnover or €20 million. The AI Act, meanwhile, requires high-risk AI systems and general-purpose AI models to disclose energy use and carbon footprint, though the first interim assessment report on the relevant efficiency standards won’t come out until 2028. Actual implementation has already revealed a real gap between rules and practice: among data centers required to disclose energy efficiency data, only 36% have actually done so, and the first mandatory reporting cycle turned up plenty of basic data errors. Overall, the EU’s legislative direction is still tightening, but the political dynamics of 2026 show that pressure for “simplification and competitiveness” is partially offsetting how binding this framework was originally meant to be.

The U.S. shows a much clearer pattern of “federal retreat, state advance,” and that split only hardened further in 2026. The SEC’s climate disclosure rule, adopted in 2024, never actually included a mandatory Scope 3 disclosure requirement even in its final form. In May 2026, the commission formally proposed scrapping the rule entirely, arguing it exceeded the agency’s statutory authority and imposed excessive compliance costs. The repeal proposal is projected to save affected companies roughly $4.9 billion a year in compliance costs, with the public comment period closing in August 2026 and a final vote possible before the year’s end. California, meanwhile, is pushing in the opposite direction with two state laws still moving forward. The Climate Corporate Data Accountability Act (SB 253) already requires large companies to disclose Scope 1 and 2 emissions starting in August 2026, with Scope 3 disclosure following in 2027. The Climate-Related Financial Risk Act (SB 261) was briefly put on hold by a temporary stay from the Ninth Circuit Court of Appeals, but companies are still broadly advised to stay “disclosure-ready” at all times. This federal-state split is, in practice, creating a “match the strictest standard” compliance logic: multinational companies, trying to avoid juggling different rules across different jurisdictions, tend to adopt the strictest applicable standard as their internal baseline and then pass it down through their supplier base. That’s the mechanism behind the so-called “California effect,” which is partially offsetting the pullback happening at the federal level.

China’s approach leans more toward a “policy plus technology” dual-track model, and it’s produced relatively verifiable results at the data center efficiency level. An action plan issued by the National Development and Reform Commission and other agencies calls for national data center rack utilization to reach at least 60% by the end of 2025, with average power usage effectiveness (PUE) dropping below 1.5. The “East Data, West Computing” initiative, which redirects data center construction toward clean-energy-rich regions in western China, has already brought the average PUE at national hub node data center clusters down to around 1.3, with the most advanced projects reaching as low as 1.04. Cities like Beijing have started charging differentiated electricity rates for existing data centers that exceed local efficiency standards. At the semiconductor level, companies including SMIC, Shanghai Simgui, and SICC have released ESG reports disclosing Scope 1 and 2 emissions data, though some are still “establishing a baseline” for Scope 3, which suggests that even in a policy environment pushing hard for disclosure, Scope 3 accounting remains a widespread weak point across the industry. China’s semiconductor equipment market now accounts for roughly 36% of global sales, the largest in the world by a clear margin. This means the pressure and opportunity around decarbonizing the domestic semiconductor supply chain are both substantial. China’s carbon market methodology framework is also expanding, adding new carbon sink categories like mangrove restoration and cold-region afforestation, which opens up space for aligning domestic carbon markets with international carbon removal certification frameworks down the line.

International organizations play a role that’s more about standard-setting and data disclosure than direct enforcement. The International Energy Agency continues to track the relationship between data center power demand and carbon emissions, projecting that data center emissions will peak around 2030 before declining slightly. The agency has also cited research suggesting that by 2035, the emissions reductions enabled by widespread adoption of existing AI applications could be three to five times larger than data centers’ own emissions, a claim tech companies frequently cite in their own marketing, though the underlying scenario assumptions and how well they hold up in practice still haven’t seen much independent verification. The OECD has called for building scenario models specifically for AI-related material demand to fill the current gap where forecasting exists for energy consumption but not for material consumption. Global South countries, which are in practice where most of these critical minerals actually come from, remain absent from international AI governance discussions. Some researchers have called for formally incorporating critical minerals into AI governance frameworks, rather than treating them as a secondary issue under supply chain security. The African Union’s related strategy has also started emphasizing local processing, value retention, and environmental sustainability, an early attempt by resource-rich countries to claim a bigger seat at the table.

The Limits of Scope 3 Accounting Methodology

Even when companies are willing to disclose, Scope 3 data often doesn’t hold up well under closer scrutiny, and there’s a structural, methodological reason for that. More than 60% of companies admit that the primary obstacle to Scope 3 accounting is a combination of poor internal data quality and uncooperative suppliers. The most common workaround is using the spend-based method, multiplying purchase amounts by industry-average emission factors instead of measuring real activity data directly. This approach is fully compliant under current rules, but the margin of error can run anywhere from 40% to 70% in either direction, which people in the industry half-jokingly call “being wrong within the rules.” There’s also a subtler tactic worth naming: call it “boundary gerrymandering,” borrowing the term from U.S. electoral district manipulation. Companies selectively report across the fifteen Scope 3 categories, deliberately excluding or understating the ones that are high-emission but difficult to get good data on, and simply chalk it up to “data unavailability.”

This black-box effect is especially pronounced in AI hardware. The Scope 3 categories most relevant to chip manufacturing (purchased goods and services, capital equipment, upstream transportation and distribution, and use of sold products) happen to be the hardest to account for accurately. Raw materials pass through multiple cross-border handoffs on their way from mine to fab, and each handoff means more decay and distortion in the carbon data attached to it. Emissions from a product’s “use phase” depend heavily on the carbon intensity of whatever grid the end user happens to be on, which makes precise cross-border accounting nearly impossible. It’s telling that it took the Semiconductor Climate Consortium a year and a half just to draft a unified accounting guideline for Category 11 alone. What’s worth watching even more closely is the strength of ESG assurance itself: most Scope 3 data today only gets “limited assurance” from third-party reviewers, not the “reasonable assurance” standard used in financial audits. Limited assurance involves a much shallower level of scrutiny, leaves far more room for error, and, if something turns out to be wrong, comes with nowhere near the accountability mechanisms that financial misstatement would trigger. Given this structural bind, the Scope 3 figures companies publish look less like the output of rigorous calculation and more like a strategic performance: an attempt to look as presentable as possible while staying technically within the rules. This aligns with the earlier discussed research on supply chain greenwashing: when the accounting method allows for such flexibility, selective disclosure naturally finds a foothold.

Can the Technology That Created the Problem Also Solve It?

Here’s a thought worth sitting with: the technology that’s made AI supply chain emissions so massive and hard to track, machine learning, large-scale data processing, and automated analytics, is also, by most industry accounts, the most promising tool for cracking open the Scope 3 black box.

Traditional supply chain carbon accounting has relied heavily on manually collecting and organizing supplier questionnaires and reports, a process plagued by inconsistent formats, delayed updates, and frequent errors. Using natural language processing and knowledge graphs, AI systems are now being used to automatically “read” and connect unstructured data like supplier emissions reports and bills of materials, describing emission sources within a unified framework and processing them at scale. With deep learning and reinforcement learning, AI can also build emissions assessment models tailored to specific supply chain scenarios quickly, using algorithms to model the tangled mess of raw material inputs and process parameters across upstream and downstream companies, replacing the old, inefficient approach of relying on manual site inspections. Blockchain technology fills a different gap: putting emissions and transaction data on-chain essentially embeds a tamper-proof, fully transparent carbon ledger into the supply chain itself. For a product like a chip, which passes through mines, smelters, fabs, packaging and testing facilities, and final assembly plants, this kind of traceability could, in theory, go a long way toward addressing the “data gets distorted with every handoff” problem discussed earlier.

That said, it’s worth being honest about where these tools actually stand: mostly still at the pilot stage, not yet proven at scale on a supply chain as complex, as sensitive, and as geopolitically contested as AI hardware. Technology isn’t a cure-all: supply chain carbon governance is, at its core, a question of development philosophy and business model. No matter how sophisticated the accounting system gets, if a company has no real intention of embedding environmental governance into its business decisions, that system just becomes one more layer of compliance theater. AI can make data collection and accounting more efficient, but it can’t replace mandatory disclosure requirements at the institutional level, and it certainly can’t replace the choice a company has to make about whether to put its core, but less flattering, suppliers on the table at all.

An Underestimated Causal Chain, and a Long-Overdue Full-Lifecycle Ledger

Back to the question we started with: what does the AI supply chain’s hidden carbon footprint actually mean?

It means that when people discuss how much electricity a single AI conversation uses (a median text prompt costs around 0.24 watt-hours, less than nine seconds of TV), that figure may well be accurate, but it only answers the last question in the operational phase. The GPU powering that conversation has already carried, from ore smelting through chip fabrication, a burden of embedded carbon and toxic materials far exceeding anything that happens once the chip is actually running. It means that when a tech company announces its data centers have achieved “carbon neutrality,” that claim may rest on numbers offset through market-based accounting rather than reflecting the real load the grid is actually carrying. And it means that when an ESG report devotes page after page to green supplier case studies, what’s quietly left out is usually the core suppliers responsible for most of the emissions, the ones that never make it into the spotlight.

The Semiconductor Climate Consortium’s push for standardized methodology; the green supplier agreements led by TSMC and Infineon; the mandatory disclosure frameworks built by the EU’s CBAM and Green Claims Directive; and international standards requiring a minimum share of primary data. All of this points in the same direction: moving corporate carbon ledgers away from voluntary, inconsistent self-reporting and toward mandatory compliance and full-lifecycle accounting. But that path is still long, and it isn’t moving in only one direction. The EU’s substantive rollback of CSDDD in 2026 and the U.S. SEC’s proposal to fully repeal its climate disclosure rule both show that this governance battle swings back and forth with political cycles. Environmental and human rights governance in mineral-producing regions still lacks any international oversight mechanism that keeps pace with how fast the AI industry is expanding. The global flow of e-waste continues to slip through the structural gaps in Basel Convention enforcement. And the selective disclosure pattern common in corporate carbon reporting, putting on a show up front while keeping the pollution backstage, still isn’t subject to anything close to the audit rigor or legal accountability applied to financial misstatement.

None of these problems are unique to the AI industry. But the sheer speed of AI’s expansion is forcing structural tensions that used to be easy to ignore or defer onto the table, with an urgency the industry hasn’t faced before. When a fab can ramp up production on a quarterly timescale and a GPU generation cycles out in eighteen months, the window regulators, rating agencies, and third-party verifiers have to catch up keeps shrinking. Procurement documents now ask for carbon footprint data. Suppliers have to sign emissions-reduction commitments. Third-party verification is becoming a condition of doing business at all. These shifts are already happening, in real terms, across the semiconductor supply chain. They just haven’t yet reached the two weakest links: mineral extraction at one end and end-of-life recycling at the other. Transparency may well be the scarcest resource in this entire green transition. And whether that transparency can stretch all the way from the fab floor to the mine pit and the teardown workshop will decide what this AI revolution ultimately leaves behind: a report card that can withstand scrutiny, or just another well-polished piece of PR.

Reference

1. TSMC, 2024 Sustainability Report / Climate and Nature Report, esg.tsmc.com

2. NVIDIA, Sustainability Report Fiscal Year 2025, images.nvidia.com；Greenpeace East Asia, “Nvidia’s Green Illusion”, 2026

3. Google, 2025 Environmental Report, sustainability.google

4. Microsoft, 2025 Environmental Sustainability Report, blogs.microsoft.com

5. Apple, Environmental Progress Report 2025/2026, apple.com/environment；Greenpeace East Asia

6. Samsung Electronics, 2025 Sustainability Report, samsung.com/global/sustainability

7. IEA, Energy and AI（2025）/ Key Questions on Energy and AI（2026）/ Electricity 2026

8. OECD, Critical Minerals and Clean Energy Applications, Environment Working Papers No. 273, 2026；OECD Due Diligence Guidance for Responsible Supply Chains of Minerals from Conflict-Affected and High-Risk Areas

9. Okolo, C. T. et al., “Global majority countries must embed critical minerals into AI governance,” Science, 2026

10. Communications Earth & Environment, “From computation to environmental cost: the resource burden of artificial intelligence,” 2026

11. ScienceDirect, “The carbon and water footprints of data centers and what this could mean for artificial intelligence,” 2025

12. PolicyReview.info, “Not greenwashing, but still… A closer look at big tech’s 2025 sustainability reports,” 2025

13. ScienceDirect, “Recalibrating global artificial intelligence e-waste estimates,” 2026

14. UN, Global E-Waste Monitor 2024；Basel Convention Secretariat, Basel Ban Amendment

15. White & Case, “Energy efficiency requirements under the EU AI Act,” 2026；European Commission, Corporate Sustainability Due Diligence

16. SEC, Rescission of Climate-Related Disclosure Rules, Federal Register, 2026；EcoVadis, “SEC Climate Disclosure Rule: Current Status,” 2026

But the reality is quite different, the truly critical resource is water. Countries like Saudi Arabia, the United Arab Emirates, Kuwait, Qatar, and Bahrain depend almost entirely on desalination to supply their cities. In some cases, more than 70–90% of drinking water comes from desalination plants.

This makes these facilities extremely sensitive infrastructure.

Recent attacks on energy facilities and the possibility of attacks on desalination plants reveal an uncomfortable truth: when access to water depends on a few giant infrastructures, the entire system becomes vulnerable.

Centralization

Centralized systems promise economies of scale and state control. But they also concentrate risk. A single facility can supply water to millions of people. This means that a single outage can leave an entire city without water.

The logic is similar to that of large electrical or telecommunications systems: the more concentrated the infrastructure, the greater the impact of its failure.

If the attacks on the Gulf’s water infrastructure reveal anything, it is the enormous fragility of centralized systems. When drinking water depends on a few giant desalination plants connected to extensive networks, the entire system becomes a strategic target.

But there is a technological and institutional alternative that rarely appears in the public debate, Decentralization.

Decentralized, closed-loop water systems.

Instead of relying exclusively on large plants that supply entire cities, each building, residential complex, or urban district could integrate its own local water system.

Such a model can include three basic components:

1. Local wastewater treatment plants (WWTPs).

Wastewater generated by the buildings themselves can be treated on-site using compact and modular technologies. This reduces the need to transport large volumes of water over long distances and allows for the recovery of valuable resources.

2. Constructed Wetlands.

Constructed wetlands function as biological systems that remove nutrients, organic matter, and pollutants through natural processes. In addition to improving water quality, they create green spaces, reduce urban temperatures, and contribute to biodiversity, even in desert environments.

3. Water Treatment and Reuse Systems.

After biological and natural treatment, water can undergo polishing and purification processes that allow for its reuse for multiple purposes: irrigation, sanitation, industrial cooling, or even human consumption when the design permits.

The result is a closed water cycle: the resource is treated, reused, and reintegrated within the same urban system instead of being lost to the sea or massive drainage networks.

Making Resilient Cities

In Gulf cities like Dubai, Doha, and Riyadh, where dependence on desalination is almost total, incorporating closed-loop systems in buildings could radically transform water security.

If each building or urban district could treat and reuse a significant portion of its water, the pressure on desalination plants would decrease considerably.

This would have several important consequences:

• Reduced strategic vulnerability. The disruption of a desalination plant would no longer lead to the total collapse of the urban water supply.

• Greater energy efficiency. Transporting and pumping water over long distances consumes enormous amounts of energy. Local systems reduce this demand.

• Environmental regeneration. Wetlands and natural systems can reintegrate water into the land instead of releasing it as waste.

• Institutional flexibility. Decentralized solutions allow multiple stakeholders, businesses, communities, and real estate developers, to participate in water management. In other words, it shifts from a model of centralized dependence to one of distributed resilience.

Water polycentrism

This approach is not just a matter of engineering. It also implies a transformation in how water is governed.

For decades, water management has been organized under highly centralized schemes, where the state or large operators control the main infrastructure.

The closed-loop model introduces a different logic: polycentrism.

In a polycentric system, multiple decision-making centers, buildings, districts, communities, and businesses, manage resources in a coordinated yet autonomous manner. This creates redundancy, institutional diversity, and adaptability.

The economist Elinor Ostrom demonstrated that this type of arrangement can be more effective at managing complex common resources than purely centralized systems.

In the case of urban water, polycentrism translates into something very concrete: many small systems instead of a few giant ones.

The Gulf Scenario

The Gulf petro-monarchies possess some of the most advanced desalination systems in the world. However, their water security continues to depend on a highly concentrated infrastructure.

Paradoxically, the same cities that lead in architectural and technological innovation could also become pioneers of decentralized closed-loop urban systems.

Buildings capable of treating their own water, districts with integrated wetlands, local reuse networks, and multiple supply sources could transform the current model.

Not only for environmental reasons, but also for strategic security.

Conclusion

The wars of the 21st century are revealing an uncomfortable truth: centralized infrastructure can become a society’s weakest point.

Water is no exception.

Faced with this challenge, decentralized closed-loop systems offer an alternative that combines technology, resilience, and institutional freedom.

It’s not about completely replacing large infrastructures, but about complementing them with a diverse network of local solutions.

Because in an uncertain world, security doesn’t come from the concentration of power, but from the distribution of the capacity to act.

Spanish

El agua como arma: lo que la guerra en el Golfo revela sobre la fragilidad de los sistemas centralizados

Durante décadas el mundo ha visto el petróleo como el recurso estratégico del Golfo Pérsico.

Pero la realidad es otra: el recurso verdaderamente crítico es el agua. Países como Arabia Saudita, Emiratos Árabes Unidos, Kuwait, Qatar y Bahréin dependen casi totalmente de la desalinización para abastecer a sus ciudades. En algunos casos más del 70–90% del agua potable proviene de plantas desalinizadoras.

Esto convierte a estas instalaciones en infraestructura extremadamente sensible.

Los ataques recientes contra instalaciones energéticas y la posibilidad de ataques contra plantas de desalinización muestran una verdad incómoda: cuando el acceso al agua depende de unas pocas infraestructuras gigantes, el sistema entero se vuelve vulnerable.

Centralización: eficiencia aparente, fragilidad real
Los sistemas centralizados prometen economías de escala y control estatal. Pero también concentran el riesgo. Una sola instalación puede abastecer a millones de personas. Eso significa que una sola interrupción puede dejar sin agua a una ciudad entera.

La lógica es similar a la de los grandes sistemas eléctricos o de telecomunicaciones: cuanto más concentrada está la infraestructura, mayor es el impacto de su falla.

Si los ataques a la infraestructura hídrica del Golfo revelan algo, es la enorme fragilidad de los sistemas centralizados. Cuando el agua potable depende de unas pocas plantas gigantes de desalinización conectadas a redes extensas, el sistema completo se convierte en un objetivo estratégico.

Pero existe una alternativa tecnológica e institucional que rara vez aparece en el debate público: los sistemas descentralizados de ciclo cerrado de agua.

En lugar de depender exclusivamente de grandes plantas que abastecen ciudades enteras, cada edificio, complejo residencial o distrito urbano podría integrar su propio sistema hídrico local.

Un modelo de este tipo puede incluir tres componentes básicos:

1. Plantas locales de tratamiento de aguas residuales (PTAR).
Las aguas residuales generadas por los propios edificios pueden tratarse en el sitio mediante tecnologías compactas y modulares. Esto reduce la necesidad de transportar grandes volúmenes de agua a largas distancias y permite recuperar recursos valiosos.

2. Humedales construidos.
Los humedales artificiales funcionan como sistemas biológicos que eliminan nutrientes, materia orgánica y contaminantes mediante procesos naturales. Además de mejorar la calidad del agua, generan espacios verdes, reducen temperaturas urbanas y aportan biodiversidad incluso en entornos desérticos.

3. Sistemas de potabilización y reúso.
Después del tratamiento biológico y natural, el agua puede pasar por procesos de pulimiento y potabilización que permiten su reutilización para múltiples usos: riego, sanitarios, enfriamiento industrial o incluso consumo humano cuando el diseño lo permite.

El resultado es un ciclo cerrado de agua: el recurso se trata, se reutiliza y vuelve a integrarse dentro del mismo sistema urbano en lugar de perderse en el mar o en redes de drenaje masivas.

De ciudades dependientes a ciudades resilientes
En ciudades del Golfo como Dubái, Doha o Riad, donde la dependencia de la desalinización es casi total, la incorporación de sistemas de ciclo cerrado en edificios podría transformar radicalmente la seguridad hídrica.

Si cada edificio o distrito urbano pudiera tratar y reutilizar una parte significativa de su agua, la presión sobre las plantas desalinizadoras disminuiría de forma considerable.

Esto tendría varias consecuencias importantes:
Reducción de la vulnerabilidad estratégica. La interrupción de una planta desalinizadora ya no implicaría el colapso total del suministro urbano.

Mayor eficiencia energética. Transportar y bombear agua a grandes distancias consume enormes cantidades de energía. Los sistemas locales reducen esa demanda.

Regeneración ambiental. Los humedales y sistemas naturales pueden reintegrar agua al territorio en lugar de expulsarla como residuo.

Flexibilidad institucional. Las soluciones descentralizadas permiten que múltiples actores, empresas, comunidades, desarrolladores inmobiliarios, participen en la gestión del agua.
En otras palabras, se pasa de un modelo de dependencia centralizada a uno de resiliencia distribuida.

Policentrismo hídrico: la lógica institucional detrás de la tecnología
Este enfoque no es solo una cuestión de ingeniería. También implica una transformación en la forma de gobernar el agua.

Durante décadas, la gestión hídrica se ha organizado bajo esquemas altamente centralizados, donde el Estado o grandes operadores controlan la infraestructura principal.

El modelo de ciclos cerrados introduce una lógica distinta: policentrismo.

En un sistema policéntrico, múltiples centros de decisión, edificios, distritos, comunidades, empresas, gestionan recursos de manera coordinada pero autónoma. Esto crea redundancia, diversidad institucional y capacidad de adaptación.

La economista Elinor Ostrom demostró que este tipo de arreglos puede ser más eficaz para gestionar recursos comunes complejos que los sistemas puramente centralizados.

En el caso del agua urbana, el policentrismo se traduce en algo muy concreto: muchos sistemas pequeños en lugar de unos pocos gigantes.

La paradoja del Golfo
Las petromonarquías del Golfo poseen algunos de los sistemas de desalinización más avanzados del mundo. Sin embargo, su seguridad hídrica sigue dependiendo de una infraestructura altamente concentrada.

Paradójicamente, las mismas ciudades que lideran en innovación arquitectónica y tecnológica podrían convertirse también en pioneras de sistemas urbanos de ciclo cerrado descentralizado.

Edificios capaces de tratar su propia agua, distritos con humedales integrados, redes de reúso local y múltiples fuentes de abastecimiento podrían transformar el modelo actual.
No solo por razones ambientales, sino por seguridad estratégica.

Conclusión

Las guerras del siglo XXI están revelando una verdad incómoda: la infraestructura centralizada puede convertirse en el punto más débil de una sociedad.

El agua no es la excepción.
Frente a este desafío, los sistemas de ciclo cerrado descentralizado ofrecen una alternativa que combina tecnología, resiliencia y libertad institucional.

No se trata de reemplazar completamente las grandes infraestructuras, sino de complementarlas con una red diversa de soluciones locales.

Porque en un mundo incierto, la seguridad no proviene de la concentración del poder, sino de la distribución de la capacidad de actuar.

This article was originally published in Spanish here.

For decades, much of the economy has followed a linear pattern: resources are extracted, turned into products, used and then discarded. This approach puts pressure on natural resources and creates growing volumes of waste.

A circular economy offers a different way of managing resources. It seeks to keep products and materials useful for as long as possible through practices such as repairing, reusing, refurbishing, remanufacturing and recycling. Organic materials can also be returned to productive use through composting and other forms of recovery.

For Nigeria, this transition could have benefits that extend beyond waste management. Circular economy activities can create jobs, support new businesses, encourage innovation, reduce pressure on natural resources and contribute to economic diversification.

Nigeria is still at an early stage of this transition. However, government policies, private-sector initiatives and international partnerships are creating some of the foundations needed for a more circular economy.

Circular practices have existed in Nigeria for generations

The idea of a circular economy may be relatively new in international policy discussions, but many Nigerian communities have long used practices that keep products and materials in use.

Household goods have traditionally been repaired rather than immediately replaced. Containers have been reused, scrap metals collected and clothing and shoes repaired. Agricultural waste has also been composted, while informal waste collectors have recovered materials that still have value.

These practices were generally driven by economic necessity rather than environmental policy. Nevertheless, they demonstrate that extending the useful life of products and recovering materials are not unfamiliar concepts in Nigeria.

The challenge is to build systems that can apply these principles at a much larger scale.

From waste disposal to resource recovery

Rapid urbanisation between the 1980s and 2000s increased the amount of waste generated in major Nigerian cities, including Lagos, Abuja, Kano, Port Harcourt and Ibadan.

Government efforts during this period largely focused on collecting and disposing of waste. Open dumping, burning and poorly managed landfills were common methods of disposal. Resource recovery and environmental sustainability received less policy attention.

As cities continued to grow, however, the limitations of this approach became increasingly apparent. Waste management requires more than simply moving waste from one place to another. It also involves deciding which materials can be recovered, reused, recycled or converted into other useful products.

This shift in thinking has gradually brought circular economy principles into discussions about Nigeria’s future development.

Global sustainability efforts have influenced Nigeria’s transition

Nigeria’s circular economy discussions have also been influenced by international developments.

The Sustainable Development Goals, the Paris Climate Agreement, the African Union’s Agenda 2063, the United Nations Environment Programme and the African Circular Economy Alliance have helped place resource efficiency and sustainable production higher on the policy agenda.

International development organisations have supported initiatives involving waste management, recycling, renewable energy and green jobs. The European Union, UNDP, GIZ, the World Bank and the African Development Bank are among the development partners identified in the source document.

Private companies have also begun investing in areas such as plastic recycling, electronic waste recovery and sustainable packaging.

Together, these developments have helped move the discussion beyond conventional waste disposal towards questions about how resources can be retained and used more efficiently within the economy.

What is Nigeria’s circular economy policy landscape?

Nigeria does not yet have a comprehensive national circular economy law. However, several existing policies and regulations support different parts of the circular economy.

Solid waste management

Nigeria’s National Policy on Solid Waste Management promotes waste reduction, recycling, composting, improved waste collection and public-private partnerships.

These measures provide some of the basic building blocks for a circular system. Better collection makes it easier to recover materials, while recycling and composting can reduce the amount of waste sent for disposal.

Environmental regulations

The National Environmental Standards and Regulations Enforcement Agency, or NESREA, administers regulations covering areas such as plastic waste, electronic waste, hazardous waste and industrial pollution.

These regulations support more environmentally sound approaches to waste management and provide a regulatory framework for several important parts of the circular economy.

Climate policy

Nigeria’s Climate Change Act of 2021 promotes low-carbon development, sustainable production, green investment and resource efficiency.

Circular economy practices can support these objectives by reducing the need for new resource extraction, improving the efficiency with which materials are used and creating opportunities for lower-waste forms of production.

Extended Producer Responsibility

Extended Producer Responsibility, or EPR, is another important part of Nigeria’s emerging circular economy framework.

Under EPR, manufacturers have greater responsibility for what happens to products and packaging after they have been used. This can encourage companies to collect used products, recycle packaging, reduce waste and consider environmental impacts when designing products.

The Food and Beverage Recycling Alliance is identified in the source document as a leading industry initiative supporting plastic collection and recycling.

Expanding EPR could become particularly important for sectors that generate large volumes of waste, including packaging, electronics and other consumer products.

Waste and renewable energy

Waste can also become a source of energy. Waste-to-energy projects can generate electricity, reduce the amount of waste sent to landfills and produce biogas.

This creates a connection between waste management and the clean energy transition. Developing such systems requires appropriate infrastructure and technology, but the potential role of waste as a resource is becoming increasingly relevant.

Plastic waste

Nigeria has also introduced measures aimed at reducing plastic pollution. These include improving collection and recycling, increasing public awareness and encouraging alternatives to single-use plastics.

Plastic is particularly important because it remains a significant waste challenge while also being a material that can potentially be recovered and reused when appropriate systems are available.

Which sectors could drive Nigeria’s circular economy?

A circular economy would affect many parts of the economy rather than a single sector. Several areas already have clear opportunities for resource recovery and more efficient production.

Waste management

Nigeria generates more than 30 million tonnes of municipal solid waste annually, according to the source document.

This creates opportunities for recycling, composting and energy recovery. Better collection and sorting would make it easier to separate materials before they are mixed together and sent for disposal.

A stronger waste management system could therefore support both environmental objectives and economic activity.

Agriculture

Agriculture produces large amounts of organic residues. Instead of treating these materials as waste, they can be used to produce organic fertiliser, animal feed, biochar, biogas and renewable energy.

This can create additional value from agricultural production while reducing the amount of organic material that is discarded.

Manufacturing

Manufacturing companies can adopt several circular approaches, including cleaner production, resource efficiency, product redesign and material recovery.

Industrial symbiosis can also create opportunities for one company’s waste or by-products to become useful inputs for another company.

For example, industrial processes can be designed to exchange materials, energy or other resources rather than treating every output as waste.

Construction

Construction generates large quantities of materials that can potentially be recovered.

Construction waste can be processed into aggregates, pavement materials, bricks and concrete products. Increasing the recovery of these materials could reduce the demand for some new raw materials while also reducing construction waste.

Electronic waste

Electronic waste presents another opportunity because discarded electronics contain valuable materials.

Proper recycling can recover materials such as gold, copper, aluminium and rare earth metals. Recovering these materials can reduce the need to extract additional resources while creating economic opportunities around collection, processing and recycling.

What is holding Nigeria’s circular economy back?

The transition faces several practical challenges. These challenges are connected, which means progress in one area often depends on improvements in others.

Fragmented policy implementation

Environmental responsibilities are shared across multiple government agencies. This can make implementation fragmented and make coordination more difficult.

A circular economy requires different parts of government to work together because waste, industry, energy, agriculture, infrastructure, finance and environmental policy are closely connected.

Clear institutional responsibilities and stronger coordination would therefore be important.

Gaps in waste infrastructure

Many communities still lack adequate collection systems, sorting facilities, recycling centres and engineered landfills.

Without these systems, even households and businesses that want to separate and recover materials may have limited options.

Infrastructure is therefore one of the basic requirements for scaling circular economy practices.

Limited public awareness

Circular economy systems also depend on how people use and dispose of products.

Many households remain unfamiliar with waste separation, recycling and circular consumption practices. Public awareness can help people understand how to separate materials, reduce waste and make better use of products.

Education therefore needs to accompany investment in physical infrastructure.

The informal recycling sector

Thousands of waste pickers already contribute to recycling in Nigeria. They recover materials that might otherwise end up in dumpsites or landfills.

However, many operate without formal recognition, social protection or access to finance.

Bringing this workforce into formal recycling systems could improve working conditions while preserving the knowledge and collection networks that already exist. The source document recommends registration, training, better occupational safety and access to finance as part of this process.

Access to finance

Circular businesses can face high operating costs and limited access to capital. They may also have difficulty finding investment incentives suited to their business models.

Recycling facilities, resource recovery systems and other circular businesses often require upfront investment before they can generate returns.

Increasing access to green finance, innovation funding and other investment mechanisms could help address this barrier.

Technology gaps

Nigeria also needs greater investment in recycling technologies, digital waste tracking, resource recovery systems and waste-to-energy facilities.

Digital tools could improve the ability to track materials and waste, while better processing technologies could increase the amount of material that can be recovered.

What opportunities could a circular economy create?

The potential benefits extend across employment, economic development, climate action and investment.

Jobs

Circular economy activities could create jobs in recycling, waste collection, repair services, green manufacturing, renewable energy and environmental consulting.

Repair and recycling are particularly relevant because they can create employment around activities that already exist at community level.

Developing these activities into formal businesses and value chains could expand their economic contribution.

Economic diversification

Nigeria’s economy has significant dependence on oil revenues. Circular economy activities could contribute to diversification through recycling industries, green entrepreneurship, sustainable agriculture and development of the bioeconomy.

The bioeconomy can create value from biological resources and agricultural materials, including through products such as organic fertiliser, biochar, animal feed and bioenergy.

Climate action

Circular practices can contribute to climate mitigation by reducing greenhouse gas emissions, methane emissions from landfills, resource extraction and energy consumption.

Keeping materials in use for longer can reduce the need for new extraction and production, while better waste management can reduce emissions associated with uncontrolled disposal.

Investment

International climate finance is increasingly supporting areas such as green infrastructure, recycling businesses, sustainable manufacturing and circular innovation.

For Nigeria, this could create opportunities to attract investment into infrastructure and businesses that support resource recovery and more efficient production.

Private-sector innovation

Nigerian startups are already developing solutions in areas such as plastic recycling, digital waste collection, waste-to-energy, circular fashion and sustainable packaging.

These businesses demonstrate that circular economy activity can extend beyond traditional waste management. Digital platforms, new business models and product innovation can all contribute to the transition.

What can Nigeria learn from other countries?

Nigeria does not need to copy another country’s circular economy model. Different countries have different economic structures, institutions, infrastructure and social conditions.

However, international experience can provide useful examples.

The Netherlands has developed approaches involving circular procurement, industrial symbiosis and material passports. These approaches focus on how governments and businesses purchase, exchange and track materials.

Finland provides an example of a national circular economy roadmap combined with green innovation funding.

Rwanda has introduced plastic bag restrictions and developed clean city initiatives.

South Africa has developed producer responsibility regulations and recycling cooperatives.

These experiences can provide ideas for Nigeria, while policies and programmes still need to reflect local economic and social conditions.

What would a stronger circular economy strategy look like?

Nigeria’s next phase of development will require coordinated action across government, business, communities, universities and civil society. The source document identifies eight areas where policy action could help accelerate the transition.

1. Develop a national circular economy strategy

Nigeria could establish a comprehensive national strategy with clear targets, defined institutional responsibilities, financing mechanisms and systems for monitoring progress.

A national strategy could help connect the different policies and initiatives that currently address individual parts of the circular economy.

2. Expand Extended Producer Responsibility

Producer responsibility could be strengthened and expanded to additional sectors, including electronics, batteries, packaging, textiles and construction materials.

This would give producers a greater role in managing products and materials after their use and could encourage better product design and recovery systems.

3. Formalise the informal recycling sector

Waste pickers should be recognised as participants in the recycling system.

The recommended measures include registering waste pickers, providing training, improving occupational safety, facilitating access to finance and integrating cooperatives into municipal waste systems.

This approach could improve working conditions while making better use of existing recycling networks.

4. Increase access to green finance

Financial institutions can support circular businesses through green loans, tax incentives, innovation grants and circular economy investment funds.

Better access to finance would be particularly important for businesses that need to invest in collection systems, recycling equipment, processing facilities and new technologies.

5. Use public procurement to create demand

Government is a major buyer of goods and services. Public procurement can therefore influence what kinds of products businesses produce.

The source document recommends prioritising products that are recycled, repairable, energy efficient and environmentally certified.

This could create stronger markets for circular products and encourage businesses to change how they design and manufacture goods.

6. Strengthen education and research

Universities can help develop the skills and knowledge needed for the transition by introducing courses in circular economy, sustainable production, industrial ecology and green entrepreneurship.

Research institutions can contribute by developing new solutions and supporting evidence-based policy development.

Education is also important outside universities. Public understanding of waste separation, recycling and sustainable consumption will influence how effectively circular systems work.

7. Invest in waste infrastructure

Nigeria needs investment in the physical systems required to collect, sort and process materials.

Priority areas identified in the document include Material Recovery Facilities, recycling plants, composting facilities, waste-to-energy plants and smart waste collection systems.

Without adequate infrastructure, many circular economy policies will remain difficult to implement at scale.

8. Encourage digital innovation

Digital technologies can help make waste and resource systems more efficient.

Artificial intelligence, the Internet of Things, blockchain and Geographic Information Systems can support waste tracking, resource management and recycling.

These technologies could improve information about where materials are generated, where they go and where they can potentially be recovered or reused.

Ending Note

Nigeria has several of the ingredients needed to begin a larger circular economy transition. Traditional repair, reuse and informal recycling practices already exist. Government policies increasingly address waste, plastics, climate change and producer responsibility. Businesses and startups are developing new approaches, while international partners are supporting related initiatives.

At the same time, major gaps remain.

Policy implementation is fragmented. Waste infrastructure is inadequate in many communities. Public awareness needs to improve. Informal recyclers need greater recognition and protection. Circular businesses face financial and technological constraints.

Addressing these issues will require cooperation across different parts of society. Government has a role in setting policies, building infrastructure and creating incentives. Businesses can redesign products, improve resource efficiency and develop new circular business models. Financial institutions can support investment. Universities and research organisations can develop skills and solutions. Communities can contribute through better waste separation, reuse and consumption practices.

Nigeria’s circular economy transition will therefore depend on how effectively these different efforts are connected.

The opportunity is substantial. Better resource management can support recycling and recovery, create employment, encourage new businesses, reduce waste and contribute to climate action. With stronger coordination, investment and implementation, circular economy practices could become part of Nigeria’s broader approach to sustainable industrialisation, economic diversification and long-term development.

References

• Ellen MacArthur Foundation, Towards the Circular Economy

• United Nations Environment Programme (UNEP), Global Waste Management Outlook

• Federal Ministry of Environment, Nigeria

• National Environmental Standards and Regulations Enforcement Agency (NESREA)

• Climate Change Act of Nigeria (2021)

• African Circular Economy Alliance (ACEA)

• World Bank, What a Waste 2.0

• African Development Bank, African Circular Economy Facility

• United Nations Development Programme (UNDP), Circular Economy and Green Growth publications

• United Nations Sustainable Development Goals, Goals 11, 12, 13 and 15

India’s climate debate often separates emission reduction from carbon dioxide removal. Reduction focuses on cleaner power, process efficiency, low-carbon transport and lower process emissions. Removal includes biomass storage, biochar, mineralization or carbonation, enhanced mineral weathering and engineered systems that enable atmospheric CO₂ drawdown.

This distinction looks essential for climate accounting, but implementation often depends on shared industrial resources: waste heat, captured gases, biomass residues, alkaline by-products, renewable electricity, water, laboratories, logistics and product markets. Carbon moves through these systems in multiple forms and across organisational boundaries, making consistent accounting and decision-making difficult. An integrated system must therefore be developed progressively to support a credible low-carbon and carbon-removal transition, and industrial symbiosis can provide this missing operational link.

This systems perspective is already reflected in India’s climate-policy direction. India’s Long-Term Low-Carbon Development Strategy calls for energy and resource efficiency, material recycling, circularity, industrial process transformation and the exploration of CO₂-removal and related engineering solutions [2]. India’s First Biennial Transparency Report (BTR) also identifies industrial networks for sharing waste heat, by-products and resources as a pathway for decarbonising hard-to-abate sectors.

India should now move beyond parallel reduction and removal efforts and build industrial systems in which both can support each other physically while remaining distinct in accounting and climate claims.

National emissions arithmetic sets the scale

To identify the real interventions required and in scales, the national emission profile must be followed wisely. In 2022, national emissions were 3,396.37 MtCO₂e, excluding land use, land-use change and forestry (LULUCF). After accounting for a net LULUCF sink of 571.82 MtCO₂e, net national emissions stood at 2,824.55 MtCO₂e. The land sink therefore offset 16.84% of gross emissions, leaving 83.16% in the national balance [1]. Although this sink represents net carbon uptake by the land sector, it should not be treated as equivalent to additional or credited carbon dioxide removal (CDR).

Between 2005 and 2022, gross emissions increased by 1,395.08 MtCO₂e, while the LULUCF sink expanded by 365.80 MtCO₂e. Thus, the terrestrial sink expansion offset 26.22% of gross-emissions growth, while the remaining 73.78% was not offset by the additional sink. This shows that net land-sector removals are nationally significant, but they cannot compensate for continued growth in gross emissions. Additional CDR must therefore complement deep emission reductions rather than substitute for them.

The composition of India’s emissions reinforces this conclusion. In 2022, the energy sector accounted for 73.37% of national emissions, with electricity generation alone contributing 40.73%. The 11 official key source categories together represented 85.32% of national emissions, indicating that India’s emissions challenge is concentrated within a relatively small number of major energy, industrial and agricultural systems.

This concentration creates an opportunity for coordinated intervention. Rather than treating each facility as an isolated boundary, emissions across connected systems of energy, materials, infrastructure and markets must be addressed. This is where industrial symbiosis becomes relevant: it can link emission-reduction measures with the physical foundations needed for future carbon-removal pathways.

Concentrated emissions require networked solutions

A limited number of source categories dominate India’s national GHG inventory, but they require different interventions. The power sector needs cleaner generation and stronger grids; transport requires lower emission systems; steel and cement require energy, material and process innovation; and biological methane and nitrous oxide require targeted agricultural management. This concentration calls for strategic prioritization, but not reliance on isolated technologies or sector-specific solutions.

Industrial symbiosis offers a way to address this heterogeneity by connecting complementary assets across sectors. Waste heat can displace fuel use in neighbouring facilities. Shared renewable power, water treatment, laboratories, CO₂ handling and logistics can reduce duplicated infrastructure. Steel slag, fly ash and kiln dust can substitute for virgin materials or serve as feedstocks for mineralization systems. Biomass residues can support biochar and biogenic-carbon pathways, while construction and highway markets can absorb verified secondary and carbonated materials derived from alkaline industrial by-products.

India’s BTR already records the targeted use of more than 63,100,000 tonnes of recycled materials, including fly ash, plastic and reclaimed asphalt, in national highway development [1]. This demonstrates the scale of demand that can emerge when standards, procurement and logistics are aligned. However, the use of a by-product is not automatically low carbon. Grinding, drying, transport, treatment and quality control must all be included in lifecycle assessment and GHG accounting.

One network can support both reduction and removal

An industrial cluster may contain concentrated biogenic CO₂ from fermentation, alkaline residues from steel or cement, renewable electricity, shared monitoring capacity and nearby construction markets [3]. Connecting these assets can create reduction and removal pathways that no single facility could operate independently.

The origin of the CO₂ determines the climate outcome. Fossil CO₂ that is captured and durably stored represents an emission reduction relative to its release. Qualifying biogenic or atmospheric CO₂ that is captured and durably stored may constitute carbon removal after additionality, durability and lifecycle emissions are accounted for. The same mineralized product can therefore serve a different climate function depending on the source of the carbon and the project baseline.

Industrial symbiosis can also combine energy substitution, avoided disposal, product substitution and durable carbon storage within the same network. The objective should be to improve the regional carbon balance, while keeping each reduction and removal outcome separately quantified. This requires the network to be deliberately engineered and properly governed.

A viable exchange requires five forms of compatibility:

• Material: composition, contamination, reactivity, particle size, moisture and variability.

• Temporal: compatible supply and demand, supported by storage and contingency planning.

• Spatial: transport distances that preserve a net lifecycle benefit by reducing leakages.

• Infrastructure: suitable pipelines, storage, preprocessing, safety and monitoring systems.

• Commercial: contracts covering quantity, quality, interruption, liability and environmental attributes.

Data are the core infrastructure of both industrial symbiosis and carbon dioxide removal. Every material, energy and CO₂ flow requires a mass and energy balance, quality specification, temporal profile, lifecycle boundary and clearly defined baseline/counterfactual scenario. Without these controls, the network may shift pollution, increase transport emissions, prolong dependence on fossil processes or create competing climate claims. The double counting and claim of climate benefits is thus a hidden challenge and risk.

Preventing double counting must be part of the design

Industrial symbiosis can generate several climate outcomes within the same network, including lower energy use, avoided waste treatment, substitution of virgin materials, reduced process emissions and durable carbon storage. These outcomes may arise from the same physical flows, but they are not interchangeable and must be quantified separately.

Because the distinction between emission reduction and carbon removal depends on the carbon source, project baseline and lifecycle boundary, these attributes must remain traceable throughout the network. Fossil CO₂ captured from an industrial point source and durably stored represents an emission reduction relative to its release, while qualifying biogenic or atmospheric CO₂ that is captured and durably stored may constitute carbon removal after biomass sourcing, additionality, the counterfactual, durability and lifecycle emissions are accounted for.

A cluster-level carbon ledger should therefore be established before operations begin. Each material, energy and CO₂ flow should have a unique identifier, defined baseline, lifecycle boundary, carbon-source classification and agreed ownership of any environmental attribute. Energy savings, avoided emissions, product-substitution benefits, gross carbon storage, lifecycle emissions and net removal should be reported separately rather than combined into a single climate total.

Overlap across Scope 1, Scope 2 and Scope 3 inventories may not be automatically double counting, because the same physical activity may appear within different organisational boundaries. Double counting arises when the same quantified reduction or removal is issued, retired or presented more than once as an exclusive claim.

The ledger should therefore identify which participant reports the operational reduction, which may report a value-chain effect and who owns any transferable carbon credit. Several organisations may contribute to the same outcome, but the quantified environmental attribute must have clear ownership. Where credits are issued, each credit should correspond to one registry record, one owner and one retirement.

India’s carbon market can finance networks—but only with disciplined accounting

India’s Carbon Credit Trading Scheme provides pathways for emission reduction, avoidance and removal. Of the 12 offset methodologies released so far, 10 address reduction or avoidance and two address biological removal; with none currently related to engineered or geochemical removal. This does not indicate the relative mitigation potential of these pathways, but it reveals a readiness gap for emerging, network-based carbon-management systems.

Industrial-symbiosis projects must be created collectively. A residue producer may provide the reactive material, a CO₂ supplier the carbon source, an infrastructure provider the transport or processing system, a project developer the technology and monitoring, and a product user the final market. Although a quantified reduction or removal must have clearly defined ownership, the wider project value can be distributed among the parties that make the outcome possible.

This distribution may occur through feedstock payments, avoided-disposal savings, CO₂-supply agreements, infrastructure charges, product revenues, long-term offtake contracts and pre-agreed sharing of carbon-credit income. Participants may also report their operational or value-chain contributions in GHG and ESG disclosures, provided that ownership of the same quantified climate outcome is not claimed more than once.

A suitable framework would therefore combine single ownership of the carbon attribute with transparent sharing of project benefits. Cluster agreements would define each participant’s contribution, responsibility, liability, revenue entitlement and reporting right before crediting begins. The registry would preserve the integrity of the credited tone, while commercial contracts would distribute the associated economic value across the wider network.

For bankability, each credited tons should retain one baseline, one climate classification, one monitoring chain, one owner and one final claim. However, the income and operational benefits may be shared across the cluster according to documented contributions and risks. In this way, interconnected industrial systems could be financed without double counting the climate outcome or excluding essential project partners from its value, and a sustainable net-zero transition can be achieved through shared responsibility, values and benefits.

References

1. Ministry of Environment, Forest and Climate Change. India: First Biennial Transparency Report to the United Nations Framework Convention on Climate Change. Government of India, 2025.

2. Ministry of Environment, Forest and Climate Change. India’s Long-Term Low-Carbon Development Strategy. Government of India, 2022.

3. Banerjee, A. Sustainability and Policy Implications for Biorefineries: Our Common Future, with Uncommon Aspirations. Doctoral thesis, 2025. https://shodhganga.inflibnet.ac.in/bitstream/10603/654049/14/14_chapter%2010.pdf

What AI Means for Developing Countries

Whether AI can help developing countries in the long run depends on three main things:

Capabilities

AI can do complicated tasks that require non-routine cognitive skills, like diagnosing diseases, predicting the weather, and giving farming advice, which are usually done by professionals who are hard to find. Developing countries have had a shortage of doctors, agricultural experts, business consultants, and public service officials for a long time. These are the kinds of skills that usually take generations of schooling to build up. The real promise of AI is that it could make expert knowledge available to everyone at a low cost.

Concentration

One example of this dynamic can be seen in the way a software developer in São Paulo writes code using American programming tools, makes reports using French software, and puts apps on cloud servers. This is an example of a supply chain that spans multiple continents. There are five parts that make up the AI stack: applications, models, data, computing infrastructure, and semiconductor hardware.

For example, a farmer in Brazil takes a picture of a sick maize leaf on his phone and sends it to an app that can diagnose the problem. The basic model may have been trained in the US using data from corn-growing areas around the world. The chips are designed in the US and made in Taiwan using Dutch lithography machines. The minerals that go into the chips are most likely from China. In a special section of the report, the raw materials used to make chips are broken down. Silicon is the main ingredient in semiconductor wafers, which are then “doped” by adding elements like arsenic, gallium, indium, and phosphorus to change their conductivity. Each element is a separate part of the global mineral supply chain.

The parts that need the most money—training cutting edge models, building huge data centers, and making advanced chips—are mostly concentrated in a few economies. Just a few big AI companies are expected to spend more on infrastructure by 2026 than the whole GDP of many countries. This focus creates risks of dependence, but it also has a pleasant surprise: latecomers don’t have to spend billions of dollars teaching frontier models everything they know because they already exist. This is because they can cut costs by a huge amount by making local changes based on open-source models.

Complementary Conditions

AI depends a lot on training data, but most of the data used to train AI systems today doesn’t really show how things are in developing countries. As an example, look at Nigeria. A medical AI tool that was trained on data from high-income countries consistently suggested too many lab tests because the clinical routines it learned worked better in Los Angeles than in Lagos.

In the 2016 World Development Report: Digital Dividends, the idea of “analogue complements” was first mentioned. This framework is based on that idea. Today’s AI needs both analogue and digital foundations to reach its full potential. In developing countries, both types of foundations are still generally weak. These include electricity, connectivity, education, and institutions; computing power; data; competition policy; and talent.

Taken together, these three variables reveal an odd truth: the smarter an AI model is, the more it needs conditions that work well with it. But the countries that need AI the most to make up for their lack of skilled workers are the ones that don’t have those conditions. AI isn’t a plug-and-play technology; instead, it’s like a magnifying glass that speeds up and shows off a country’s existing infrastructure problems, weak data foundations, and institutional flaws at speeds that have never been seen before. So, the very idea of a “AI opportunity” should be taken with a grain of salt, since the existence of such an opportunity depends on how well a country prepares its foundations before AI comes along.

Economic, Social, and Political Impacts

A whole part of the report looks at both sides of AI’s effects on developing countries. This fits well with ESG analytical frameworks, which put the effects into three groups: social services, political power structures, and economic output.

Economic Dimension

The report shows a big difference: in high-income economies, 14.2% of jobs could be automated by generative AI, but only 4.5% of jobs in low- and middle-income economies. At the same time, 16.2% of jobs could have their skills “augmented” instead of being replaced. For now, it looks like AI will have less of an immediate effect on job markets in developing countries than on those in high-income countries.

This is not good news, though, because it shows a different kind of lag. Businesses in developing countries are much slower to adopt AI than businesses in wealthy countries. This is especially true for complex applications like workflow automation, where large companies with lots of resources are still the main ones adopting AI.

As an example of a successful case study, Telangana, India, has an AI-powered weather forecasting system that helped smallholder farmers change their planting choices based on specific local risks. This helped some farmers increase their agricultural spending by one-third, while others saved up to $560 per household. On the other hand, countries that depend on call centers and back-office Business Process Outsourcing (BPO) are under a lot of competitive pressure from AI automation. This is because machines are taking over tasks that humans used to do for cheap.

Social Dimension

There are two main areas where AI could be useful in public services: areas where there aren’t enough professionals (like medical imaging diagnosis) and areas where centralised systems can make prediction more accurate (like weather forecasting). The report talks a lot about how successful Kenya’s court system is. Each year, AI assigns more than 10,000 cases to more than 1,500 mediators, which helps clear up long-standing backlogs.

But the risks are just as real. If training data doesn’t include local people, languages, and cultures, AI results could become very biased. They can also be easily used as weapons to spread false information, commit fraud, and commit cybercrime. These threats are amplified in countries that don’t have strong regulatory systems.

Political Dimension

The most tense part of the report is the section on politics. It is divided into four axes of power that AI has changed: power between nations, power between governments and corporations, power between businesses and individuals, and power between citizens and the state.

Controlling important parts of the value chain, like chips, data centers, cloud networks, and essential minerals, gives a country power. The US has been limiting the export of advanced semiconductor chips since 2018. In response, China has limited technologies and elements used in rare-earth processing. All of the parts of the AI tech stack could become a problem in geopolitical games. Both the U.S. “AI Export Initiative,” which began in 2025, and China’s “Digital Silk Road,” which has been going since 2015, are trying to get technological footholds in developing countries by offering money and access to markets. The report’s trade data shows that most of Latin America’s exports go to the United States. On the other hand, East Asian and Pacific countries depend on both superpowers in different ways.

Between the people and the government: The report uses a typical example to show this dual potential: an AI camera system that was first put in place to help with traffic and speed up emergency response can also be used to identify and keep track of protesters, turning it from a public safety tool to a surveillance system. You can’t just let AI decide what happens in politics; who controls the technology, sets the rules, and has the power to challenge them is what really matters.

This part of the report is probably the most interesting and the least tech-optimistic. The political dimension brings the real issues to the forefront, while the economic and social narratives stay mostly focused on the question of “can AI help?” AI is never neutral infrastructure; it has political code built into it. People who control the upstream are in charge of making the rules.

“Whose AI is being used, and under what conditions?” is a much more useful question for developing countries than “should AI be used at all?” This is the reason why environmental and social factors alone are not enough to look at AI through an ESG lens. It’s easy to forget about the governance pillar, especially when it comes to cross-border power structures, but it is the single most important thing that determines long-term outcomes.

Hidden Costs - Environment and Mind

Besides the main text, the report has a number of special sections. Two of these deserve particular attention because they go against the common story of “what AI can do” and ask a tougher question: who is paying the hidden costs of AI?

The Cost to the Environment

In terms of the environment, the report says that AI has big and possibly skyrocketing environmental costs throughout its whole lifecycle, from extracting important minerals to running data centers and getting rid of electronic waste. Developing countries are especially at risk in this area.

To illustrate the scale, generative AI is expected to make 5 million metric tonnes of electronic waste by 2030, up from 1.2 million metric tonnes in 2020. That’s more than 6% of all the electronic waste in the world. At the same time, only 1% of all the demand for critical rare-earth minerals is being met by recycling.

The report says that if there aren’t enough rules, corporations that are willing to risk the health of local communities in exchange for short-term profits will have even more negotiating power. This is a big red flag for developing countries that already have trouble enforcing the law.

The Cost to the Mind

Several studies cited in the report show that relying too much on generative AI can hurt cognitive abilities and make people less motivated to learn on their own.

One study that looked at Chinese students in grades 7 through 12 found that as the use of AI rose from almost nothing to about 80%, students’ monthly test scores went down by 20%, even though their homework scores went up. Students who spent the same amount of time doing assignments on their own, on the other hand, did much worse on tests.

Over 80% of high school teachers surveyed across 51 high-, middle-, and low-income countries were worried that AI could give bad or inappropriate advice, and 65% were worried that AI could make students’ cognitive biases stronger. It’s interesting that teachers in low- and middle-income countries were less worried about these issues. The report suggests that this might be because people in those areas aren’t as used to AI tools as others. However, it doesn’t answer the troubling question of whether or not students in these developing countries are really able to use AI tools critically.

The three-stage ‘Adopt-Adapt-Advance’ pathway

The suggestion step-by-step is clearly progressive: Adapt, Advance, Adopt. Each preceding stage requires far less investment, skill, and infrastructure than the one that follows.

Advance - Frontier models, independent research

The biggest obstacle to entry is advance, or the independent development of frontier models. It requires special chips, huge data centers, mountains of training data and top-notch researchers, something affordable today only to a few nations and corporations.

The most active pioneers in this space include India’s Sarvam AI, which has gone from tweaking existing architectures to building novel models; Africa’s Lelapa AI, which created InkubaLM, a model trained directly on five local African languages; and the United Arab Emirates, which developed the Arabic-focused Falcon model series. The upward trend reflects the evolution of the semiconductor industry. Malaysia has built up decades of experience in chip assembly and manufacturing. In 2025, local firm SkyeChip released its first self-designed 7-nanometer chip, the MARS1000.

Adapt - Localization with Open Source Models

Adapt – creating local solutions on top of open-source models – is the most cost-effective route. Entrepreneurs can use open-source frameworks like Apertus, BERT, Gemma, Llama, Mistral and Qwen to build applications that are finely tuned to local languages, legal systems and domestic needs. Examples include Latam-GPT in Latin America and the SEA-LION language-model network in Southeast Asia. Many of these local adaptations are better than even the biggest global models for regional languages.

Adopt - Use the Tools Already Available

The simplest way to start is “adopt”, i.e. use off-the-shelf AI tools directly. The value of AI is only dictated by frontier models with the largest parameter counts. “Small AI” refers to models with smaller parameter footprints or narrow task-specific designs, built for environments with unstable power grids and limited connectivity, capable of running offline directly on mobile phones. Let’s look at some practical use cases:

Nuru from Kenya: Works offline on mobile phones to better identify crop leaf diseases than human agricultural extension officers.

Cough Against TB: Developed by Wadhwani AI in partnership with India’s Central Tuberculosis Division, this tool accurately detects nine out of ten patients suspected of having tuberculosis who need further testing, based only on audio recordings of coughs.

PROMPTS Chatbot: A randomised controlled trial in 40 healthcare facilities in Kenya found that mothers who interacted with the chatbot sought medical care for themselves and their children much more frequently.

Rori Project, Ghana: Offers AI-powered math tutoring through basic SMS. With poor cellular signals, twice-weekly 30-minute sessions enabled students to make progress equivalent to a year’s learning in eight months for around $5 per person.

Unacknowledged Realities and Constraints

Some two structural constraints that are often neglected and deserve separate attention are:

The Connectivity Gap: 3.4 billion people worldwide still lack access to mobile internet. Of these, 3.1 billion are in areas with network coverage but are still offline due to prohibitive pricing, and 300 million have no signal coverage at all. Ironically, the very rural and low-income communities that could benefit most from AI are the ones that telecom operators deem as commercially unviable.

Economic Rationality vs Technological Capability: AI applications are not always adopted even when complementary conditions exist. Computer-vision-based “just-walk-out” retail automation is commonplace in high-income countries but rare in low- and middle-income countries where local cashier wages are a tiny fraction of the cost of purchasing, installing and maintaining sensors and software. Likewise, automated garment-cutting machinery has existed for decades, but big apparel exporters like Bangladesh, Cambodia and Vietnam seldom embrace them. Lack of complementary conditions is not always the main barrier to AI adoption. Often the decision not to adopt AI is a rational economic response to low labour costs in the local economy.

The “Adopt-Adapt-Advance” pathway sounds like a growth strategy, but it is really a risk management approach, not a growth goal. In many national policy discussions, “indigenous AI R&D” is used as a symbol of national prestige, as if not reaching the Advance stage is tantamount to falling behind.

But the cases of retail automation and clothing manufacturing suggest a more mundane reality: that the availability of technology is not the same as economic rationality, and cost structures are the final arbiter. For the vast majority of developing countries, the unreachable Advance tier is far less practical than learning how to develop truly cost-effective localised applications in the Adapt tier. This method is more realistic, and it also gives real results much faster.

Roles of the Governments

To deal with AI’s effects on the economy, society, and politics, governments need to play three separate roles at the same time:

As a Helper/Facilitator

Governments need to build both analogue and digital foundations. Analogue foundations include affordable electricity, connectivity, devices, and basic skills. Digital foundations include computing power, open data, competition policies, and talent. To help local businesses adapt to AI, they need to use changes to public procurement. At the same time, they need to improve social safety nets for people whose jobs or ways of making a living are lost because of AI.

Short-term AI adoption needs active information-sharing and matchmaking mechanisms to connect ready-made AI solutions with potential business users. Just having the infrastructure in place doesn’t automatically create a market.

The report uses an interesting fact to show how unequally this role is being played: by mid-2026, more than 80 countries had released national AI strategies, but only Rwanda was one of the 25 low-income countries that had done so. This gap shows that the “enabler” role is often missing just where it’s needed the most.

As a Customer/User

AI should only be used by governments in areas that really fit their administrative capabilities and data conditions. It is just as important to invest in AI technology as it is to find the right problems for it to solve. This shows a useful difference between front-end and back-end applications:

Implementing front-end AI (services that people interact with, like support tools for teachers, medical staff, and social workers) will be harder in the near future because of things like limited hardware, connectivity, and ease of use.

Back-end AI (administrative tasks like predictive analytics for tax, weather, transport, and pension departments) is usually the first type to be widely used. This is because it has a smaller user base and more digitalised data.

In fact, most of the AI used by governments in low- and middle-income countries right now is in back-end tasks like predictive analytics and decision support, remote sensing and geospatial intelligence, and environmental monitoring and early warning systems.

Better results often come from setting up systematic ways to evaluate and keep an eye on things, making administrative data “AI-ready,” preserving negotiating power over vendors by using small, flexible contracts to avoid vendor lock-in, and giving digital teams, not traditional IT purchasing departments, control of AI procurement.

As a Chief Observer/Regulator

Starting with voluntary standards to build trust, examining how current laws cover AI-related harms (like discrimination, privacy violations, and cybersecurity holes), gradually filling in regulatory gaps, and strengthening enforcement and compliance capacities at the same time should be the regulatory path.

There is no one model for regulation that works for everyone. Countries need to make their own governance frameworks by trying new things, building up their own skills, consulting the public, and being open and honest. At the same time, working together with other countries stops regulatory havens and market fragmentation. This gives developing countries a bigger say in how global rules are made.

How Roles Work Together

These three roles are not separate at all. For example, if the government puts too much pressure on its regulatory role, it can actually make it less effective as an enabler, making it harder for the private sector to come up with new AI ideas.

Attention is on improving social safety nets for people who are losing their jobs or income because of AI.

It also looks into how to build cross-departmental coordination systems to stop AI policies from being scattered within the company.

In many developing countries, going digital, protecting data, and regulating industries are all handled by different ministries that don’t work together. This makes AI governance look organised on paper but hard to put into action.

This framework requires a government with naturally limited resources to do three tasks that can’t be done at the same time: spending money to build infrastructure, being a smart, budget-conscious customer (the “buyer”), and making a regulatory framework that will work before the private sector pushes back.

This is a big job for any government, let alone one in a developing country. The role of User stands out as the most important. The other two roles can’t really take root unless the government makes good decisions about procurement and data governance. For the money put in by the enabler and the rules made by the regulator to really pay off, the government has to show that it can use them. If this level doesn’t move forward, it’s easy for the other two roles to become nothing more than paper exercises.

Ending Note

The most important thing about this report is not that it is full of hope, but that it rejects two simple stories -

1. The “AI is a silver bullet” myth: The report busts this myth with real-life examples, like how the medical AI tool didn’t work in Nigeria or how retail automation wouldn’t make economic sense in places with low labour costs.

2. The doomsaying that “developing countries will always be left behind”: The report shows that even in places with few resources, regions can still make a difference in the world by finding the right entry points, as shown by projects like Nuru in Kenya and Rori in Ghana.

In the end, how advanced the technology is never makes a difference. Instead, it depends on whether a country put money into basic “grunt work” early on, like making sure there is electricity, connectivity, good data hygiene, and talented people, and on whether its government has the administrative skills to keep the balance between encouraging innovation and reducing risk.

For people who are interested in ESG and sustainable development, the report gives them more of a framework for asking questions than a clear answer. As AI spreads quickly, who is paying the hidden costs in terms of the the environment and the mind? Who is really getting the benefits from the economy and society? How much room is there to change this pattern of distribution? These questions are much more important than asking if AI will change the world, and they should be followed up on all the time.

Reference

World Bank. 2026. World Development Report 2026: The Promise of Artificial Intelligence. World Bank.

Green skills are the knowledge, capabilities, values, and practical competencies needed to support a more sustainable and low-carbon economy. They are not limited to environmental scientists, renewable energy engineers, or sustainability managers. Increasingly, green skills are becoming relevant to finance professionals, supply chain leaders, manufacturers, educators, marketers, engineers, hospitality operators, policymakers, and entrepreneurs.

This is why sustainability education can no longer be treated as a specialist subject. It is becoming a core workforce requirement.

The Green Transition Is Reshaping Jobs

Climate change, resource scarcity, ESG regulations, circular economy expectations, carbon reporting, responsible sourcing, and clean energy investment are changing the way organizations operate. Businesses are under increasing pressure from regulators, investors, customers, employees, and communities to demonstrate responsible environmental and social performance.

As a result, the demand for people who understand sustainability is growing. The World Economic Forum’s Future of Jobs Report 2025 identifies climate-change mitigation and climate-change adaptation as major trends expected to transform business and workforce strategies. It also highlights growing demand for roles connected to renewable energy, environmental engineering, and electric and autonomous vehicle technologies. [1]

This tells us something important: sustainability is no longer only about corporate responsibility. It is becoming part of workforce planning, business strategy, innovation, and competitiveness.

Green Skills Are Needed Across Every Function

One of the biggest misunderstandings about green skills is that they belong only to “green jobs.” In reality, almost every job can become greener.

A procurement professional needs to understand responsible sourcing, supplier ESG requirements, product lifecycle impacts, and Scope 3 emissions. A finance professional may need to understand climate-related financial risks, green finance, ESG reporting, and sustainability-linked investment. A manufacturer must understand energy efficiency, waste reduction, circular production, material traceability, and carbon reduction. A marketing professional must understand green claims, consumer expectations, and the risk of greenwashing. An educator must prepare learners to think critically about sustainability challenges and solutions.

In other words, the future workforce will not only need technical skills and digital skills. It will also need sustainability intelligence.

The Green Skills Gap Is a Business Risk

The challenge is that demand for green talent is growing faster than supply. LinkedIn’s green skills research has repeatedly emphasized that workers with green skills are not increasing fast enough to meet demand, creating a gap that can affect climate action, business transformation, and workforce readiness. [2]

For businesses, this gap can slow down ESG implementation, carbon reduction programs, sustainability reporting, green product innovation, and supplier compliance. For workers, it can limit career opportunities in a labor market increasingly shaped by sustainability expectations. For countries, it can affect competitiveness in the green economy.

This is why green skills development must become a priority for schools, universities, professional bodies, employers, and training institutions.

Sustainability Education Must Be Practical

Sustainability education should not stop at awareness. Knowing that climate change is a problem is important, but it is not enough. Learners and professionals must also know how to act.

Practical sustainability education should help people understand how to measure carbon emissions, reduce waste, improve energy efficiency, evaluate suppliers, apply circular economy principles, understand ESG regulations, identify climate risks, and make responsible business decisions.

The International Labour Organization emphasizes the importance of equipping individuals with the knowledge and competencies needed for green jobs and supporting the transition to sustainable and environmentally friendly economies. [3]

The next generation needs sustainability education early. Professionals need continuous upskilling. Business leaders need the ability to connect sustainability with strategy, compliance, innovation, cost savings, and long-term value creation.

Green Skills and Digital Skills Must Work Together

The future workforce will also need to combine green skills with digital skills. Artificial intelligence, data analytics, digital product passports, carbon accounting platforms, supply chain traceability tools, and smart manufacturing systems are all becoming part of sustainability management.

This means sustainability professionals must become more data-literate, and technology professionals must become more sustainability-literate. The future will reward people who can connect environmental knowledge, business thinking, and digital tools.

The Role of Educational Institutions and Employers

Educational institutions have a responsibility to prepare students for the world they are entering, not the world that existed twenty years ago. Sustainability should be integrated into business, engineering, supply chain, finance, hospitality, aviation, technology, and management education.

UNESCO’s Greening Education Partnership aims to prepare every learner with the knowledge, skills, values, and attitudes needed to tackle climate change and promote sustainable development. It also connects education with the transition toward digital and green economies. [4]

Employers also have a responsibility to train their workforce. Sustainability cannot remain only within one department. It must become part of organizational culture, decision-making, procurement, operations, product development, reporting, and leadership.

This requires a shift from “sustainability as a topic” to “sustainability as a capability.”

Conclusion

Green skills will define the future workforce because the future economy must operate within environmental limits, social expectations, and evolving regulatory requirements. The workers and organizations that understand this shift early will be better prepared for the opportunities ahead.

Sustainability education is not just about protecting the planet. It is about preparing people for the future of work, business, leadership, and responsible innovation.

The question is no longer whether green skills are needed. The real question is whether our education systems, companies, and professional training programs are moving fast enough to develop them.

The future workforce will not only need technical skills and digital skills; it will need sustainability intelligence

References

1. World Economic Forum. The Future of Jobs Report 2025. https://www.weforum.org/publications/the-future-of-jobs-report-2025/

2. LinkedIn Economic Graph. Global Green Skills Report / Green Skills Resources. https://economicgraph.linkedin.com/research/green-skills-resources

3. International Labour Organization. Skills for Green Jobs. https://www.ilo.org/topics-and-sectors/skills-and-lifelong-learning/skills-green-jobs

4. UNESCO. Greening Education Partnership. https://www.unesco.org/en/sustainable-development/education/greening-future

Why Did Medicine Become a Linear System?

The development of a new medicine follows a familiar sequence. A molecule with therapeutic potential is identified, tested, refined, scaled, and eventually delivered to a patient. At each stage, the focus is almost entirely on whether the process works clinically: is it effective, is it safe, can it be made at scale. The resources consumed along the way and the waste generated are largely treated as operational overhead rather than as design variables. A molecule that fails a toxicity screen is not repurposed; it is abandoned. A delivery system optimized for one drug is rarely carried over to another. Each development pathway moves forward toward a single therapeutic goal, with little consideration for what happens to the materials and processes left behind.

What changed is scale. Estimates suggest that drug discovery alone generates up to 2 million kilograms of waste per year, with a further 1.5 million kilograms produced during preclinical studies, before a compound ever enters a human trial. Individually, milligram-scale laboratory syntheses may appear insignificant. Collectively, however, thousands of such experiments conducted simultaneously across the globe generate an environmental footprint that is far from trivial.

Drug development follows a funnel, in which many candidates are screened and only a handful progress to manufacturing. Environmental scrutiny generally begins only once kilogram-scale production is underway, by which point the core chemistry has already been selected and optimized, making redesign both difficult and expensive.

What Does “Waste” Actually Mean in a Formulation Laboratory?

Inside a formulation laboratory, waste is not simply what reaches a disposal container. It represents the resources invested in developing candidates that ultimately do not progress. Consider what happens during the development of a nanoparticle-based drug delivery system: many candidate formulations are screened; varying polymer types, lipid compositions, particle sizes, and surface coatings to find the combination that meets the clinical target. Most candidates fail, leaving behind discarded reagents and solvents, along with the energy consumed to synthesize, purify, and characterize compounds that never progress beyond development.

This challenge is amplified by solvent selection. Data from over 400,000 profiled reactions in pharmaceutical discovery chemistry showed that more than 40% were conducted in solvents classified as undesirable, with methylene chloride and N,N-dimethylformamide appearing most frequently. These solvents remain common because they are reliable and well-established, and replacing them during an active development program can introduce significant technical and financial barriers.

In pharmaceutical development, clinical performance determines whether a formulation succeeds or fails. The environmental burden incurred during its development receives comparatively little weight, even when hazardous waste generation is substantial. This asymmetry reveals a fundamental limitation in current design priorities.

There is, however, a promising technical shift underway. High-throughput experimentation (HTE) enables researchers to miniaturize and parallelize reactions, allowing hundreds of conditions to be tested simultaneously using small quantities of material. Beyond accelerating reaction optimization, HTE generates structured datasets that can train machine learning models to predict successful chemical conditions before extensive experimental testing begins. Applied to formulation development, these approaches could expand the search for safer and more sustainable alternatives without increasing material consumption. Instead of relying on established solvents or conditions because they have worked previously, researchers can evaluate a wider range of possibilities and identify patterns that would otherwise remain hidden. Although these tools are not yet standard practice across pharmaceutical research, they demonstrate how the same computational infrastructure being developed to accelerate discovery could also reduce waste at its earliest source.

Why Isn’t Recycling Enough?

Recycling often feels like the obvious solution: recover solvents, treat waste streams, and dispose of hazardous materials more responsibly. These measures are important, but they address waste only after it has already been created.

More than 25 years ago, Paul Anastas and John Warner placed waste prevention at the forefront of the twelve principles of green chemistry. Their first principle is simple: design chemical processes to avoid generating waste rather than finding better ways to manage it afterward. Recycling a solvent is preferable to incinerating it, but both approaches still assume the solvent was needed in the first place. Eliminating or replacing it during process design removes the problem before it exists.

This distinction determines where meaningful change can occur. Solvent selection, reaction pathways, and formulation strategies are established early in development, when there is still flexibility to make different choices. Once a process has been optimized, validated, and scaled, changing it becomes considerably more difficult. The greatest opportunity for reducing pharmaceutical waste therefore lies upstream, during the earliest stages of research and development.

What Happens When Design Leads?

The benefits of upstream design decisions are already evident in industrial pharmaceutical processes.

Pfizer and sertraline (Zoloft), 2002: The original manufacturing process for sertraline, one of the world’s most widely used antidepressants, involved a three-step synthesis requiring four solvents (methylene chloride, tetrahydrofuran, toluene, and hexane). Pfizer redesigned the process into a more efficient single-step route using a selective catalyst that reduced impurity formation and minimized the need for reprocessing. By applying green chemistry principles, Pfizer doubled overall yield, reduced raw material use by 20-60% and eliminated approximately 1.8 million pounds of hazardous materials. The redesign was not driven primarily by environmental ambition. It was driven by the recognition that a cleaner process is also cheaper and more efficient, and that the two goals pursued together reinforce each other.

Merck and sitagliptin (Januvia), 2006 and 2010. The synthesis of sitagliptin, a treatment for type 2 diabetes, was redesigned twice in a short period - first through a novel catalytic route, then again through a fully biocatalytic process developed in collaboration with Codexis. The final process reduced waste by over 80%, completely eliminated aqueous waste streams, and created 220 pounds less waste for every pound of sitagliptin manufactured. Over the lifetime of the drug, Merck projected the elimination of more than 330 million pounds of waste. The key change was replacing a conventional chemical step with an enzyme, a biocatalyst engineered specifically for this reaction, biodegradable, renewable, and capable of performing the transformation under far milder conditions than its chemical equivalent.

These examples highlight an important distinction. The waste reduction did not come from adding a treatment step after an established process. It came from revisiting the design of the process itself and asking, at the earliest stage, whether the chemistry could be built differently.

What Would This Look Like Earlier in Development?

The success of the Pfizer and Merck redesigns also reveals a limitation in current practice. Both involved drugs that were already approved, already at scale, already generating waste at industrial volume before anyone asked whether the process could be built differently. The improvements were substantial, but they were corrective measures applied after the original design choices had already shaped years of manufacturing.

The gap between when a design decision is made and when its consequences become visible enough to prompt action is where most of the damage accumulates. Early formulation and synthesis decisions often prioritize performance, reliability, and familiarity. Solvents are selected because they dissolve the compound effectively or because a laboratory already has experience using them. Material choices are guided by therapeutic function, while questions about degradation, persistence, or environmental impact are rarely considered at the same stage.

A design-led approach would incorporate these factors from the beginning: selecting synthetic routes that minimize hazardous solvents, evaluating delivery materials according to their environmental fate as well as their therapeutic function, and integrating recovery strategies before waste becomes a manufacturing problem.

The tools to support these decisions already exist. Solvent selection guides, green chemistry frameworks, and biodegradability databases provide researchers with the information needed to make more sustainable choices. What remains missing is the expectation that environmental considerations should influence therapeutic design from the earliest stages rather than being addressed after development is complete.

Design is only one part of the story. Once a medicine leaves the laboratory, it enters a much larger system of manufacturing, regulation, clinical practice, and resource management, where another set of environmental questions begins to emerge. Those questions will be the focus of the next article in this series.

To understand how Global brands are adapting, sustainability in FMCG can be categorised into three pillars.

Sustainable Sourcing

Sustainability begins long before a product ever hits a factory assembly line. It starts when a product’s environmental impact is decided by the choices made at the very beginning – where its raw materials come from. The FMCG products we buy on daily basis rely heavily on inputs like palm oil, cocoa, soy and paper. Actually, in the past sourcing these materials at a large scale have caused deforestation and biodiversity loss.
Today when we mention sustainable sourcing it simply means carrying out strict tracking systems to ensure raw materials do not contribute to environmental destruction.
The strategy used by corporate supply chains are shifting toward certified sourcing frameworks. For example; many brands now commit to using 100% RSPO certified (Roundtable on Sustainable Palm Oil) materials- It’s like a green stamp of approval for palm oil as it means they are buying from farmers who have officially proven that they grow their oil safely, without environmental destruction. By doing so, companies protect local ecosystems while securing their own long-term supply chains against climate risks serving as a business benefit.

Circular Packaging

The Fast Moving Consumer Goods (FMCG) industry has used the linear “take-make-waste” model for decades․ Cheap single-use plastic packaging and non-recyclable multi-layer wrappers make goods easy for consumers to consume on-the-go‚ and these products and other types of plastic packaging are then thrown away‚ ending up in landfills or the ocean ecosystem․ Redesigning these wrappers into a closed loop‚ circular economy system reduces long-term environmental and regulatory concerns that global brands are facing․ Strategy: The largest packaged consumer goods companies are rethinking product lifecycle blueprints by reducing the use of virgin (newly produced) plastic‚ lightweighting packaging‚ eliminating use of problematic plastics (such as PVC) and increasing the use of Post-Consumer Recycled (PCR) plastics (i.e. a shampoo bottle is sent to be recycled and then used for a new shampoo bottle)․
The Real-World Impact; The Ellen MacArthur Foundation’s Global Commitment Progress Report signatories taken together used three times more PCR plastic in 2022 than in 2018․ They have also reduced their use of virgin plastic by 6% overall‚ showing that companies can reduce their use of new raw materials and still grow their businesses․

Green Logistics

Even if a product contains 100% sustainable ingredients in a circular bottle‚ its score is reduced by the carbon emissions associated with transporting it from factory to shelf․ The distribution networks‚ cold storage facilities‚ and trucking fleets of consumer goods companies are a big part of the industry’s hidden emissions․ Sustainability of the consumer goods business can be realized through aggressive‚ efficiency-driven supply chain optimization around “green logistics” and operational decarbonization․
The Strategy is Manufacturers are optimizing their footprints by adding increasing amounts of renewable energy to plants‚ using AI route-optimization software to make delivery truck routes more efficient and switching their transport fleet to electric or alternative-fuel vehicles․
In its most recent corporate sustainability report‚ consumer goods company Unilever reported a 77% reduction in its Scope 1 and 2 operational GHG emissions compared to 2015 levels․ This important decrease was driven mostly by the transition to 100% renewable grid power for manufacturing facilities and factory energy efficiency improvements․This shows that a focus on and investment in operational emissions reductions can lead to speedier action․

Conclusion

Ultimately, “sustainability” in the FMCG sector is far more complex than placing a green leaf icon on a box or releasing a limited-edition eco-friendly item. It requires a connected, three-part system: sourcing ingredients without destroying nature, packaging products so materials stay out of landfills, and transporting goods with minimal carbon emissions.
As global regulations tighten and consumers demand greater accountability, companies that treat these three pillars as core business operational goals rather than short-term marketing campaigns will be the ones that build resilient, future-proof supply chains.

The question they should be asking is far more uncomfortable, and far more strategic:

Can investors place their trust in our governance?

The talk around the world has now moved on from simply producing more detailed sustainability reports or including ESG commitments on corporate websites. It has become what many refer to as the “post-ESG era”, in which organisations are assessed on what they can show through good governance, transparency, accountability, and tangible results.

It is ironic that this change should occur at exactly the time when the Caribbean has its best chance.

Although the Caribbean accounts for less than 1% of global greenhouse gas emissions, it is still one of the region’s most vulnerable to climate risks. In recent decades the annual average of climate-related losses has been about 2.1% of regional GDP, which places a huge strain on the finances at the same time as increasing the need for resilient investment.

Nevertheless, investors are still moving towards the region.

They are being more selective.

Sustainable investment on a global scale is still managing tens of trillions of dollars in assets, and development finance institutions are quickly increasing their climate and sustainability financing. At the same time, sustainability disclosure is becoming more standardised through the International Sustainability Standards Board (ISSB), establishing a common language for the capital markets even though there are regulatory differences between jurisdictions.

Capital is not any longer restricting itself to sustainable projects; it is now seeking governable ones.

• Instances in which decisions are recorded.

• Risks are quantified.

• Boards are accountable.

• Data are auditable.

• ESG performance can withstand independent scrutiny.

The Caribbean is becoming a laboratory for sustainable finance

Although international headlines tend to concentrate on Europe or North America, a number of the most innovative sustainable finance mechanisms are appearing in the Caribbean.

Barbados has developed debt-for-resilience arrangements which channel debt service into investments in both climate and social areas, thus creating a model that development banks hope to adopt in the region.

The Caribbean Development Bank has greatly speeded up its climate finance. In 2025 alone the value of its climate approvals went over US$226 million, which was about half of all the approvals it granted, and the Bank has pledged to raise its climate finance to 35% of its own resources by 2028 as well as being preparing a US$200 million regional Blue Economy Programme.

Meanwhile, regional institutions are collaborating with the International Finance Corporation (IFC) on the creation of a regional green finance taxonomy, while nations like the Dominican Republic have already established national sustainable finance frameworks which offer greater certainty to investors.

Investors are asking different questions -

1. Who is responsible for the sustainability risks at board level?

2. Do businesses include climate risks in their risk management plans?

3. How are human rights kept under review as they relate to supply chains?

4. Can the ESG metrics be verified independently?

5. Is there a connection between executive compensation and sustainability performance?

6. How fast can the organisation react to a change in regulation?

These questions do not focus on environmental performance.

They are concerned with the quality of governance.

It is remarkably difficult to copy governance.

Caribbean competitiveness has long been based on tourism, logistics, financial services and natural capital.

Over the next decade, another asset may become equally valuable:

Institutional trust.

Countries that are able to establish reliable regulatory environments, transparent capital markets, effective environmental governance and trustworthy ESG oversight are likely to draw a greater proportion of sustainable investment.

This is especially relevant given that financing within the blue economy is expanding.

Investors are now calling for governance systems which are able to deal with environmental and social risks over decades rather than just during election cycles, whether the areas in question are sustainable ports and renewable energy, coral reef restoration, resilient infrastructure, marine biotechnology or sustainable tourism.

The future belongs to countries that have attractive projects as well as those which have reliable institutions.

Looking toward 2030: governance will determine who attracts capital

In 2030 the Caribbean organisations are likely to be working in a completely different investment environment.

Climate disclosure will become an ever-greater part of financial reporting.

Supply chains will have to meet stricter due diligence rules.

Artificial intelligence will take over ESG monitoring and compliance.

Financial disclosures related to nature will serve to complement climate reporting.

Blue finance will turn into a mainstream practice rather than remaining experimental.

Any organisation that starts to improve its governance now will be in a position to take advantage of these new capital flows, while those that postpone their actions may find that financing is limited by the quality of the project as well as their governance capabilities.

The challenge at this stage for the Caribbean is to persuade investors that the institutions in the Caribbean are capable of ensuring sustainability with the same degree of credibility, transparency and accountability that is expected in the world’s foremost financial markets.

References

• Caribbean Development Bank. (2025). Climate finance investment doubled to US$226.7 million in 2025. https://www.caribank.org/newsroom/news-and-events/cdb-climate-finance-investment-doubled-usd2267-million-2025

• European Financial Reporting Advisory Group. (2023). European Sustainability Reporting Standards (ESRS).

https://www.efrag.org

• International Finance Corporation. (2023). Blue Finance: A Practical Guide for Financial Institutions.

https://www.ifc.org

• International Sustainability Standards Board. (2023). IFRS S1 General Requirements for Disclosure of Sustainability-related Financial Information.

https://www.ifrs.org

• International Sustainability Standards Board. (2023). IFRS S2 Climate-related Disclosures.

https://www.ifrs.org

• Organisation for Economic Co-operation and Development. (2026). Caribbean Development Dynamics 2026: Strengthening Resilience and Competitiveness.

https://www.oecd.org

• Reuters. (2025, July 9). Barbados to be test case in first regional debt swap scheme.

https://www.reuters.com

• Reuters. (2026, May 26). Two steps back, three forward: Sustainability reporting and global capital markets.

https://www.reuters.com

• Task Force on Climate-related Financial Disclosures. (2017). Recommendations of the Task Force on Climate-related Financial Disclosures.

https://www.fsb-tcfd.org

• United Nations Environment Programme Finance Initiative. (2023). Nature Positive Finance and Sustainable Financial Systems.

https://www.unepfi.org

• United Nations Global Compact. (2023). Forward Faster Initiative.

https://unglobalcompact.org

• World Bank. (2024). The Blue Economy Development Framework.

https://www.worldbank.org

• World Economic Forum. (2025). The Global Risks Report 2025.

https://www.weforum.org

Concerning the ESG framework, the letter “S,” which stands for “Social,” is the hardest to measure and the one that companies often forget about. A business can measure carbon in tonnes. It can give a number to the number of women on its board. But there isn’t a simple way to measure how much a worker’s diminished dignity is worth, how badly a miner gets hurt while making tech hardware, or how many job applicants an AI recruitment tool quietly leaves out.

The difference is starting to close thanks to rules about work. GRI 401, GRI 404, and GRI 405 set standards for hiring, training, and diversity. People could give their thoughts on GRI 406 starting in July 2025. The ISSB has added information about the workforce to IFRS S1 and S2. Companies must report on employment and diversity under the EU’s CSRD, ESRS S1. Regulations are not keeping up with how quickly AI is being used.

The Future of Jobs Report from the World Economic Forum, which came out in January 2025, says that AI will create 170 million jobs by 2030 while destroying 92 million, for a net gain of 78 million jobs. The net number hides a correlation that is not positive. There are new job openings in AI engineering and data science. An overwhelming majority of the lost job opportunities are in administrative, data entry, and customer service roles that used to pay good, middle-class wages.

The five parts of AI’s social cost that this article looks at are job loss, the digital gap, algorithmic bias, hidden labour, and the changing social contract.

The Reality of Employment Displacement

The Discrepancy in Labour Market Structure

In previous waves of automation, there wasn’t as much unemployment as analysts thought there would be. The difference this time is not in size, but in speed and accuracy.

In the past, automation mostly affected manual, blue-collar work. Generative AI focuses on reading, writing, reasoning, and translation, all of which are important skills for white-collar jobs. Gen-AI: Artificial Intelligence and the Future of Work, a study released by the IMF in January 2024, says that AI will affect about 40% of jobs around the world, and that number will rise to 60% in developed countries.

The IMF divides this kind of influence into two different groups. “Displacement” is the term for when AI takes over tasks, reducing the need for people to do them. When AI helps workers be more productive and create new value, this is called “augmentation.” In industrialised economies, 60% of jobs are at risk. About half of these jobs could be added, and the other half are facing real job loss or wage pressure. Emerging economies are more than 40% exposed, while low-income countries are only about 26% exposed. This is mostly because AI isn’t widely used in those labour markets.

The Future of Jobs Report 2025 from the WEF, which is based on data from over 1,000 large companies and 14 million employees, gives us more information. The jobs that are going away the fastest are postal clerks, bank tellers, data entry workers, and administrative helpers. The jobs that are growing the fastest are those that work with AI and machine learning, big data, cybersecurity, and renewable energy. About 22% of formal jobs around the world will go through structural changes by 2030. New jobs need years of training, but roles that have been around for a while can be eliminated during a single revenue cycle.

McKinsey’s longer-term view is more dire: between 2030 and 2060, about half of current jobs may be replaced, with the median estimate being around 2045, which is about ten years earlier than previous predictions. AI and digital twin technology can now run whole production lines, from design to assembly, with very little help from humans. This has a big impact on manufacturing.

Economists Daron Acemoglu and Pascual Restrepo stress that AI’s effects go beyond just taking away jobs; they also say that AI can eliminate some tasks. AI often takes on 40% to 60% of a role’s responsibilities instead of the whole job, which means that employers need fewer people to get the same work done. This kind of displacement is harder to see in big data, but it has just as big of an effect on wages and the structure of jobs.

Demographic Vulnerabilities

Displacement is not spread out evenly across the workforce. The 2025 report from the World Bank is called “Disruption Without a Dividend?” The data are put into groups based on gender, age, and skill level.

Sex

It is a fact that women are much more likely than men to work in dangerous jobs. It is 17% more likely for women to lose their jobs to automation in high-income countries than it is for men. In upper-middle-income countries, the difference is 10% instead of 6%. The structural reason for this is that jobs that are mostly held by women, like office work, customer service, and data entry, are areas where AI has made a lot of progress. Since the 1990s, these roles have helped women keep their jobs, but AI is quickly removing that safety net.

Jobs that AI is likely to add to instead of take away are more evenly split between men and women. Women lose more than their fair share in AI-driven labour markets, but profits are shared more evenly. This shows a clear pattern of gender inequality in these markets.

Age

Young employees also have to deal with the same problem. 10.4% of workers aged 16 to 35 in upper-middle-income countries are in high-risk jobs, compared to 4.3% of workers aged 56 to 65. This doesn’t make sense, since young people are usually thought to adapt the fastest. However, young people are more likely to work in entry-level jobs, which are the first jobs that companies automate. U.S. data show that the number of jobs for people aged 22 to 25 in jobs that are affected by AI has dropped by about 13% compared to jobs that are less affected by AI.

Older workers face unique problems, such as finding it harder to retrain, dealing with higher psychological stress, and losing their jobs at a time when family costs are highest. Japan’s Ministry of Health data show that only 35% of people aged 50 and up who lost their jobs were able to find new ones within two years, while 67% of people under 35 were able to do the same. This difference is likely to put stress on pension plans over the next ten years.

Proficiency

PwC’s 2025 analysis of about a billion job postings showed that workers with AI skills earn up to 56% more than workers in the same roles who don’t have those skills. On the other hand, the pay of workers without AI skills stays the same or goes down. AI isn’t just changing the number of jobs that are available. It’s making the difference in pay even bigger between people with the same job titles. The World Economic Forum (WEF) says that by 2030, the most important skills for workers will have changed by 39%. AI and big data skills are expected to grow the fastest.

Geographic Disparity and Obstacles to Retraining

There is no coexistence of new and old jobs in the same places.

The Bay Area, Beijing, London, and Singapore are some of the hubs where employment in new technology is concentrated. On the other hand, the service and administrative jobs that are being eliminated are spread throughout many smaller cities. While the losses affect real communities, national statistics show net gains. Without legislative action, this could result in regional decline similar to the U.S. Rust Belt, where the advantages of automation primarily benefited coastal areas, leaving states like Ohio, Michigan, and Pennsylvania to suffer for decades without the expected jobs.

59% of the world’s workforce will need retraining before 2030, according to WEF data. Of this group, 19% can change careers, and 29% can retrain in their current roles. However, 11% of workers—more than 120 million—are seriously at risk of being laid off, primarily because there aren’t enough opportunities for retraining for three reasons.

First, money and time. Unlike students, working adults are unable to stop working in order to pursue full-time education. Training must take into account the need to fulfil financial commitments.

Second, outdated content. Between 2024 and 2025, the demand for AI competencies increased by about 70%. Updates to training programmes are substantially behind schedule. When a course is introduced, it might already teach antiquated skills.

Thirdly, access is not equal. Reliable internet connectivity is necessary for online training, which is primarily conducted in English. This prevents older people, low-income households, rural workers, and non-English speakers from participating.

The Digital Divide

A group of numbers that go against what you think

DataReportal and We Are Social put together the Digital 2026 Mid-Year Global Update Report, which looked at more than 240,000 internet users aged 16 and up from 54 countries. The monthly use of AI around the world is 81.2%. But the countries with the most usage are not what most people think they are.

Kenya (97.5%), the UAE (94.2%), and Indonesia (93.6%) use AI the most. On average, 91.8% of people in South-east Asia use AI. Countries known for their advanced AI research have lower scores: the US has a score of 71.5%, the UK has a score of 62.8%, Germany, France, and Canada all have scores between 65% and 72%, and Japan is the only country out of the 54 that was looked at that scored below 50%. China’s score is 86.4%, which is just above the world average.

One thing to keep in mind is that the 81.2% number includes all AI-enabled features, not just active chatbot use. It includes AI search summaries and built-in assistants. Still, the difference in rankings is important and needs to be explained.

Why is AI being used more in emerging markets?

Not technology, but necessity is what drives it. In developing markets, AI gives people access to services that were out of reach for them before, like tutoring, translation, and basic legal advice. In wealthy countries, on the other hand, AI is often just one option among many high-quality services that are already available, making the switch less appealing. Loss aversion, a well-known principle in behavioural economics, makes this problem even worse: people in wealthy countries are more likely to fear the losses that AI could cause, like jobs, privacy, and independence, than to see its benefits, because they have more at stake.

The Stanford HAI’s AI Index Report 2025/2026 backs up the trend: 83% of people in China and 80% of people in Indonesia think AI is more helpful than harmful, but only 40% of people in Canada and 39% of people in the US agree. The less wealthy a country is, the more optimistic its people are; the more prosperous a country is, the less optimistic its people are.

A lot of people may not benefit from a high penetration rate. The Strange Case of the Digital Divide in Emerging Markets

High utilisation does not mean there is a big benefit. This is what the UNDP’s December 2025 report calls the main paradox: in emerging markets, AI is mostly used for free, simple consumer apps, while better uses like business automation, research, and industrial use are slowed down by bad digital infrastructure. The World Bank has numbers to back up this difference: of the 441.8 million jobs that could be helped by AI, 66.9 million do not have access to the internet to take advantage of these benefits. This means that almost 15% of the potential gains have not been realised. Also, the jobs that can be done online, like office work, data entry, and customer service, are the ones that are most likely to be automated. This is what the World Bank calls a “small buffer, big bottleneck”: low-income countries are the least resilient, but they will be the first to be affected by AI.

How the lack of infrastructure affects different parts of the world

According to UNESCO, only 14% of people in the least developed countries can connect to the internet from their homes. Just having a connection isn’t enough. To use AI effectively, you need devices that work well and, more often than not, language representations that work well. Over 90% of the data that modern AI models are based on comes from just four languages: English, Chinese, French, and Spanish. In addition to the digital divide, people who speak Swahili, Hausa, Bengali, Punjabi, and many other languages also face a linguistic one. For example, medical advice that doesn’t work across cultures is an example of an AI response that might be wrong or even harmful in another situation.

The UNDP sees this as a three-part problem: skills (AI experts are mostly concentrated in a few hubs), computing power (most GPUs are in the US and China), and governance (developing countries often don’t have the tools to control AI or protect workers). All of these things could make the differences that were seen during the Industrial Revolution worse more quickly.

The Strange Case of AI Viewpoints: Why Developed Countries Are More Hesitant Japan is different. About one in four Japanese people are worried about AI, and only one in ten think it will make things worse in the future. However, Japan has the lowest rate of adopting AI of the 54 countries that were surveyed, at below 50%. This does not mean calm confidence. It shows how three decades of inaction have made people less sure that new technology can change things. Because of a lack of workers, care robots and factory automation are now necessary rather than optional. However, people still don’t trust small steps forward as much as they do in the most sceptical Western countries.

South Korea took a different approach. According to data from the Bank of Korea, more than half of Korean employees use generative AI at work. This is about twice as common in the US and represents the fastest growth rate in the world. Concerns about AI are only voiced by 16% of South Koreans, a much smaller percentage than the over 40% of people in the U.S., Italy, and Australia. Even though they live in similar areas and face similar problems related to getting older, the outcomes are not the same. The difference is not in the technology, but in how people think the future will be better.

Algorithmic Discrimination and Workplace Surveillance

How to Hire AI

An early example is Amazon’s internal AI hiring tool, which stopped working in 2018. Since it learned from ten years of hiring data from a male-dominated field, phrases like “women’s college” or “women’s chess club” were taught to hurt resumes.

With the rise of many AI-based hiring tools, algorithmic bias went from being a theory to a legal fact. In 2024, researchers from the University of Washington sent three AI models identical resumes, with only the applicants’ names changed. 85% of the time, the models chose names that sounded white, while only 9% of the time, resumes with names that sounded black were recommended. They also consistently chose male names.

In May 2025, a federal court in California let Mobley v. Workday move forward as a class action. This was the first time that a U.S. court allowed a group lawsuit about AI hiring discrimination to go forward. Lead plaintiff Derek Mobley, a Black American man over 40 years old, said that since 2017, more than 100 companies had turned him down for jobs using Workday’s AI screening, citing his race, age, and disability as reasons. In February 2026, the court gave permission to let all applicants over 40 know that they had been turned down through Workday since September 2020. An important finding that goes beyond this case is that AI software may be considered an employer’s “agent.” This means that vendors can’t get out of liability by calling their product a “tool.”

The EEOC won its first AI-bias case in 2023, reaching a settlement with iTutorGroup. The company’s hiring programme routinely turned down women over 55 and men over 60, building age discrimination into its code.

These situations all follow the same pattern. Most of the time, bias is not encoded on purpose; instead, it comes from past data. If hiring practices were unfair in the past, AI trained on that data will continue to do so, though more quickly and with less transparency. The legal idea of “disparate impact,” which means unintentional bias that causes unequal harm, is slowly being applied to AI. However, in most countries, it’s still not clear who has to prove something and what they have to say.

The Problem of Bias in Identifying People by Their Faces and Photographs

In 2018, MIT researcher Joy Buolamwini’s Gender Shades project measured the differences in facial recognition software. They found that error rates were up to 35% for dark-skinned women and less than 1% for light-skinned men. This was because the training data wasn’t diverse enough. After these facts came out, IBM and Microsoft stopped selling the software to police departments.

But deployment has become more important than rectification. Face recognition technology is now used a lot in public safety, border control, and background checks. Because of the error disparity, Black people, people with darker skin, and women are more likely to be wrongly identified. Two different studies done in 2025 on AI-generated images showed similar biases: when there are no limits, most models show doctors as white men, nurses as women, and criminals as people with darker skin. These outputs don’t just reflect stereotypes; when looked at in a big picture way, they reinforce them.

What happens when algorithmic management hurts workers’ freedom?

Bias goes beyond hiring people. In the workplace, algorithmic oversight is becoming more and more clear. Amazon warehouse workers are watched to see how fast they pick, how much they move, and how long their breaks are. If they don’t meet certain standards, they get automatic alerts, and repeated alerts can lead to performance reviews and firings. Employees call this “robot management” because they can’t talk to a supervisor about it; there’s only a display of metrics.

Similar to ride-hailing apps, food delivery apps assign orders and pay drivers based on how well they complete tasks and how happy customers are with their service. A well-known study of China’s food delivery industry found that shorter delivery times force riders to break traffic rules and drive into oncoming traffic in order to keep their ratings. This is a safety risk that is caused by algorithmic incentives rather than individual choice.

The EU’s Platform Work Directive (2024/2831), which went into effect in October 2024, is the strongest law we have seen so far. It gives platform workers the right to know how algorithms affect their pay and working conditions. It also bans certain uses of personal data in managerial decisions. The law affects about 28.3 million platform workers in the EU, and that number is expected to rise to 43 million by 2025. It has to be written into national law by December 2, 2026, for member states to follow through.

How ESG frameworks react to discrimination based on algorithms

In July 2025, GRI started consultations on the revised GRI 405 and GRI 406, which were made to match ILO rules. The consultations asked companies to share examples of discrimination and the steps they were taking to protect vulnerable groups. Algorithmic bias is now known to be a type of systemic risk because so many AI tools are being used in HR decisions this quickly.

Local Law 144 in New York requires hiring tools to be checked for bias once a year. Similar rules are being worked on in Colorado, Illinois, and California. The EEOC and the Department of Justice have said that AI cannot shield employers from anti-discrimination laws. AI may make decisions easier, but it does not shield employers from legal responsibility.

Concealed Exploitation

The Human Framework for “Ghost Workers” Helping the Mechanism Work

Behind the safety filter of each AI chatbot is a team that looks at the worst content that users have written. Identifying data, managing content, and learning from human input through reinforcement learning are all necessary for AI systems to work. Human evaluation also helps models learn what a good response looks like. None of these tasks can be done without human help. These “ghost workers” are mostly found in Kenya, the Philippines, India, Venezuela, and Colombia. They are hired through platforms like Remotasks and Scale AI and get paid one to three dollars an hour to check for violent, sexual, and extremist content while regular users don’t see it.

In November 2025, the Institute for Human Rights and Business released a report that called content moderation “the sweatshop of the AI era.” The report blamed Meta and TikTok for not doing enough to protect these workers. It was found that the most important AI labour lawsuit in the world happened in Nairobi. Between 2023 and 2024, more than 140 Kenyan workers sued Facebook after reviewing harmful content for the company through the outsourcing company Samasource Kenya. A medical exam of 144 workers ordered by the court showed that more than 140 had been diagnosed with PTSD, anxiety, or depression after seeing pictures of murder, suicide, child abuse, and terrorism. Employees said they had to sign non-disclosure agreements and got little to no help with their mental health.

In the Philippines, India, and Colombia, the same pattern can be seen. A 2025 Equidem study of 113 digital workers in Kenya, Colombia, and Ghana found that the content they were looking at was directly linked to over 60 cases of serious mental distress, such as PTSD, depression, insomnia, and anxiety. Most of them didn’t have access to mental health support while having to deal with high production standards.

Meta paid a lot of money for lawyers. In 2020, they had to pay $52 million to settle moderator claims in the U.S., and in 2021, they had to pay $85 million to settle a case in California that affected over 10,000 moderators. Similar lawsuits have been filed against TikTok. This shows that the damage isn’t just happening in the Global South. Kenyans who used to work for Sama have sued OpenAI, saying that data labelling for ChatGPT caused them mental distress. This shows that even the biggest AI companies can be affected by bad conditions.

Things that the system does to hide harm

It’s possible that the most important thing about this job is not the low pay, but the requirement of silence. Before they start working, almost all digital employees have to sign strict non-disclosure agreements that say they can’t tell anyone, even therapists, about their work.

In Equidem’s poll in Colombia, 75 of the 105 workers who were asked to take part refused, mostly because they were afraid of breaking their non-disclosure agreements. This forced silence makes it hard to measure the damage and makes it harder to hold people accountable. It also goes against ILO Convention 190, which says that long-term psychological harm is the same thing as violence at work. Both the US and Kenya have not agreed to this treaty. The US is home to Meta, Google, Microsoft, and OpenAI. In February 2025, the Philippines passed laws like Code AI Coalition, and now Colombia’s legislature has approved rules that are similar under CONPES 4144 National AI Policy. Ratification fits in well with the places where this work is done.

A lack of regulation in digital work and the absence of ESG

This shows a problem with the way things are set up: tech companies use multi-tiered outsourcing, which treats digital workers as independent contractors. In most countries, this means they don’t get minimum wage, insurance, or safety protections by law. This shows up as almost complete silence in ESG reports. Meta, Google, Microsoft, and Amazon all give a lot of information about their direct employees but not much about the digital workers in their supply chains. Of course, GRI 408 and 409 are supposed to be about supply chains, but the rules for digital work are much weaker than the rules for physical goods.

Over 700 digital platform workers from around the world were evaluated by the Fairwork Project at Oxford. Of the 15 major platforms that were looked at, none met the basic standards for all five criteria: fair pay, good working conditions, fair contracts, fair management, and enough representation. The public gave feedback on the GRI’s new LRBR standard from December 2025 to March 2026. This is the first attempt to include these indirect links into the GRI’s framework. However, it usually takes a few years from consultation to adoption and then compliance.

Corporate Strategy Transformations

The Difference Between the Public and Experts

According to Stanford HAI’s AI Index Report 2026, there is a greater difference within the US than in the majority of other countries. There is a fifty-point difference between the percentage of experts who think AI will improve employment (73%) and the general population (23%). The economic numbers are 69% as opposed to 21%. 84% as opposed to 44% in the healthcare sector.

Just 33% of Americans think AI will improve their job prospects, which is less than the 40% global average. Private AI investment in the United States totalled $285.9 billion in 2025, which is 23 times more than China’s. The United States ranked 24th out of 54 countries surveyed, with only 71.5% of American adults using AI on a monthly basis. Investments in AI are mostly focused on corporate tools rather than products that improve the everyday lives of the average person.

Technology Redundancies and the DEI Retreat

In 2025, American tech companies’ social responsibility discourse underwent a dramatic change. Private businesses imitated the Trump administration’s decision to eliminate federal DEI programmes. About 20% of businesses had already discontinued DEI programmes by July 2025, and another 16% were expected to follow suit by the year’s end.

In November 2025, Microsoft announced that their annual diversity report would no longer be released. This was the company’s largest reduction in social transparency to date, and it also eliminated about 6,000 positions while hiring AI engineers. Meta rescinded 6,000 proposed responsibilities and eliminated about 8,000 positions, or 10% of the workforce. Snap cut 16% of its workforce. Salesforce laid off 4,000 support workers after CEO Marc Benioff made it clear that the company needed fewer employees. Oracle increased its investments in AI while disclosing large layoffs.

None of these businesses disclosed the support given to fired employees, nor did they identify which jobs or skill sets were most negatively impacted. Despite dwindling commitments to diversity and job security, Microsoft, Google, Amazon, and Meta collectively plan to invest roughly $320 billion in AI by 2025.

Global Policy Mapping

The European Union

When it comes to creating artificial intelligence regulations, the European Union has made the most progress. The EU AI Act, which was passed in 2024 and goes into effect in 2025, is the first piece of legislation in the world to regulate artificial intelligence in its entirety. It views AI as a product that requires safety evaluations as well as a potential threat to fundamental rights, with risk levels dictating the necessary standards. The legislation’s sponsor, MEP Dragos Tudorache, claims that regulation is “an opportunity to articulate our values.”

Recruiting, worker-management, and access-to-work AI tools are classified by the Act as high-risk, requiring an independent safety assessment before going on sale and ongoing supervision after. People who are affected by these systems can ask a national authority for clarification. The EU AI Office began conducting cross-border compliance assessments in 2025 with a focus on public services, employment, education, and credit. Countries in the European Union are updating their unemployment policies to require companies using AI technologies to provide at least three months of transitional support.

Approximately 28.3 million gig workers are covered by the EU’s Platform Work Directive (2024/2831). In the event that a platform is unable to prove that a person is truly self-employed, that person is categorised as an employee, with full rights, the capacity to understand algorithmic management, and a ban on platforms using private conversations for management decisions. By December 2026, member states must put the directive into effect.

The United States

The United States at the federal level. The focus of the AI labour policy has changed. The Biden administration ordered the Department of Labour to release fair employment guidelines and included worker protections in its AI executive order. Several of these actions were revoked by the Trump administration in 2025, and the EEOC has since pursued allegations of AI bias with less vigour.

The shortfall has been addressed by the states. California improved vocational coverage in low-income areas by about 20 percentage points by increasing training funding for workers most at risk from AI. Artificial intelligence businesses that create jobs are eligible for tax breaks in New Jersey. Local Law 144 in New York requires bias assessments when using AI. When artificial intelligence assesses job interviews, Illinois requires disclosure. Although its precise ramifications are still unknown, Microsoft and the AFL-CIO came to an agreement in December 2023 to inform workers about AI developments and to support reskilling. This was the first collaboration between a major technology company and organised labour. Suggestions for a robot tax or AI dividend have been sparked by MIT economists Acemoglu and Restrepo’s observations of declining employment and wages in regions with high robot usage; California and New Jersey have made some progress, but there hasn’t yet been a federal initiative.

Japan

Japan is one of the developed countries most at risk from AI-induced job displacement, according to AMRO data from 2024. However, after three decades of stagnation, the public’s response has been muted, marked more by tiredness than confidence. In contrast to the EU’s rights-based approach, which allows workers to challenge algorithmic decisions, Japan’s labour ministry requires that at least 10% of positions remain human-staffed when corporations implement AI.

Japan established a national AI Safety Institute and published AI Governance Guidelines for Businesses in April 2024. These are concrete steps, but while Japan primarily relies on guidelines and voluntary commitments, Europe enforces legally binding laws with penalties.

China

China is actively promoting the commercialisation of AI while concurrently enacting early limitations on algorithms’ labor-related capabilities. The Provisions on Algorithmic Recommendations, which forbade algorithms from unfairly discriminating in price and established essential safeguards for gig workers, including delivery riders, were passed in 2022. The public outcry over a 2021 viral post about riders risking their lives to avoid automated delivery timers prompted this.

China imposed strict regulations on generative AI content in 2023. These regulations required the labelling of AI-generated materials, including deepfakes, and required algorithms with significant social impact to be registered and have their compliance evaluated. For the first time, AI-induced job displacement was given top priority on the national agenda in September 2024 thanks to a joint policy from the Central Committee and the State Council. This policy also updated occupational standards to formally recognise new roles like AI Application Engineer and Data Labelling Specialist, giving training programmes a clear focus.

South Korea and Singapore

Singapore takes an elite-first approach: starting in 2025, every student will receive five to ten hours of AI instruction, backed by a S$120 million Science of AI programme that equips teachers to use it. The larger strategy follows the same logic: take swift and precise action, changing the labour market before disruption occurs rather than after.

Rapid adoption was the course taken by South Korea. According to data from the Bank of Korea, more than half of workers currently use generative AI in their work, which is more than twice as high as what is seen in the US. Korea became the second country after the EU to enact systematic AI regulation that included labour, safety, and transparency in a single statute when it passed a comprehensive AI law in December 2024. While Japan, after three decades of stagnation, largely lacks optimism for another economic advancement, Korea, despite living in the same region and facing similar ageing pressures, does.

The Global South and the African Union

The African Union published its Continental AI Strategy in 2024. But there is a real paradox here: Kenya has one of the highest rates of AI utilisation in the world (97.5%), but it is primarily dependent on free consumer tools and has nothing to do with domestic AI research or industry.

In three dimensions, Africa’s position is unbalanced. The rates of user adoption are among the highest in the world. As digital labourers, African workers carry out the riskiest and lowest-paid jobs necessary for AI systems to function. Due to the lack of training data for African languages and the concentration of computational resources in the US, China, and Europe, Africa is notably underrepresented in the developer community.

Important emerging economies show variations as well. Brazil is developing AI regulations, but the details are still up in the air. India is increasing access to AI by utilising its Digital Public Infrastructure. Indonesia uses AI extensively, but its infrastructure and governance are uneven and have not changed to reflect this. The lesson is that the “Global South” is not a single entity, and it will not work to replicate a governance model from Washington or Brussels in Jakarta or Nairobi. Each situation calls for a different reaction.

Universal Basic Income

The current safety net was not built for the disruption brought about by artificial intelligence, but rather for an employment framework from the industrial era. Gig and informal workers have no safety net because unemployment insurance requires a formal employer. While social security funding is based on stable, long-term employment, which AI is undermining, retraining programmes are updated too slowly to keep up with the rapid advancements in AI.

Due to this discrepancy, Universal Basic Income—a constant, unconditional cash payment to every person—has returned to the forefront of important policy discussions. Three pilots offer concrete proof. In Finland’s 2017–2018 experiment, 2,000 jobless people received an unconditional monthly stipend of 560 euros. While mental health and trust in governmental institutions improved, employment slightly increased rather than decreased. A similar pilot programme was carried out in Stockton, California, from 2019 to 2021, giving 125 people $500 per month; participants found full-time jobs faster than the control group. The GiveDirectly initiative in Kenya, which ran from 2016 to 2024, saw an increase in small businesses among participants, better nutrition, and higher levels of educational attainment. The widely held belief that free income reduces the incentive to work is refuted by all three results. The main obstacle is financial: funding a universal currency system in the face of a shrinking tax base brought on by automation poses a serious problem, which is why a robot tax or AI dividend is frequently discussed.

What Workers Require in Order to Request

Employees need rights that traditional labour laws did not foresee as artificial intelligence increasingly controls work processes: the right to know how algorithms make decisions about their employment, the right to challenge incorrect decisions, the right to access their own work data, and the right to negotiate the deployment of particular AI systems. In the US and Europe, some unions have started pushing for this. Most have not made much progress.

The Future of ESG

ESG reporting is currently disjointed. The majority of the GRI labour regulations are optional. Only within Europe does the EU’s CSRD have legal authority. Financial risk is the main focus of the ISSB. The discrepancy permits substantial oversight: in 2024, roughly 60% of large employers used AI monitoring techniques, such as video surveillance and keystroke recording, but less than one-third of monitored workers were aware of this, creating a glaring problem for worker dignity under GRI 403.

Five Essential Measures for the Artificial Intelligence Era

Openness regarding layoffs. Although GRI 401-3 requires the disclosure of transition support, some companies use phrases like “restructuring” to mask AI-induced reductions. Any organisation that reduces a particular position by 10% or more as a result of AI must identify the impacted individuals by role, age, and location, as well as the assistance received.

Algorithmic fairness audits. High-risk system audits are required by the EU AI Act; however, public disclosure procedures vary. GRI 405 and 406 should require the results of Employment-AI audits, broken down by age, gender, and race, along with a clear statement about the auditor’s independence.

Thorough coverage of the supply chain. Indirect labour within AI value chains is hardly addressed by current frameworks. In addition to disclosing median compensation, mental health benefits, and the use of non-disclosure agreements, platforms should examine their data labelling and content moderation subcontractors with the same rigour as they do physical goods vendors.

Artificial intelligence and mental health. Physical safety is the main focus of GRI 403. It should include concerns about automation, algorithmic pressure, and surveillance anxiety as valid health risks at work.

Information about platform employees. Large platforms must now provide worker statistics, average compensation, insurance coverage, and procedures for challenging algorithmic decisions, just like they would with direct employees, since the EU now automatically classifies platform workers as employees.

Conclusion

With generative AI growing quickly, AI infrastructure has become an important asset class in the global economy. As a result, the economic and environmental aspects of AI can no longer take precedence over its social effects.

There is no doubt about the evidence. In contrast to content moderators paid between $1 and $3 per hour under non-disclosure agreements, Kenya has a 97.5% AI usage rate; of 144 employees assessed by a court-appointed physician, more than 140 were diagnosed with depression or PTSD. There is a 50-point difference between the 73% of experts who believe AI will benefit the economy and the 23% of average Americans who agree, indicating a disagreement over beneficiaries rather than the capabilities of the technology. Over 60% of large companies currently use surveillance software to monitor employees without their consent, and AI recruitment tools clearly show bias.

According to the UNDP’s 2025 Human Development Report, society will determine how AI develops rather than it being a technical necessity. A useful viewpoint is offered by Amartya Sen’s Capability Approach: social policy and corporate regulations should go beyond simple market efficiency and protect people’s true freedom to preserve their health, respect their dignity, and manage their own professional lives. In order to prevent the burden of this shift from disproportionately affecting those who are least able to handle it, the social component of ESG must serve as a real check on authority.

Reference

World Economic Forum (WEF), The Future of Jobs Report 2025, January 2025.

International Monetary Fund (IMF), Generative Artificial Intelligence: The Future of Employment, January 2024.

World Bank, Disruption without Dividends? The Influence of Digital Disparities and Task Variations on the Global Effect of Generative AI, 2025.

United Nations Development Programme (UNDP), The Next Great Divergence: The Potential for AI to Exacerbate Inequality Among Nations, December 2025.

UNDP, Human Development Report 2025: A Matter of Choice — Humanity and Possibility in the Age of AI, 2025.

CNN Business, “Facebook Left Kenyan Content Moderators with Enduring Trauma,” December 2024.

The Brink, The Ghosts Within the Machine: AI’s Concealed Human Trauma, July 2025.

Global Reporting Initiative (GRI), Labor-Related Topic Standards Project, 2024-2026 (includes proposed amendments for GRI 401, 404, 405, 406, and LRBR).

Interoperability Guidance for ESRS S1 and GRI-ESRS, EFRAG, 2024-2025.

The European Union’s Artificial Intelligence Act (EU AI Act) was officially adopted in 2024.

European Union, Platform Work Directive (Directive 2024/2831), enacted in October 2024.

ASEAN+3 Macroeconomic Research Office (AMRO), Regional Evaluation of AI-Induced Employment Displacement Risks, 2024.

African Union, Continental Artificial Intelligence Strategy, 2024.

Groundwater — the water stored beneath the Earth’s surface in geological formations known as aquifers — supplies drinking water to billions of people, supports food production across continents, and maintains the flows that keep rivers and wetlands alive. In many regions, it is the primary basis of water security.

According to the UNESCO World Water Development Report (2022), groundwater accounts for approximately 99 percent of all liquid freshwater on Earth and supplies around half of all water withdrawn for domestic use globally, providing the primary source of drinking water for rural populations without access to centralised water systems. It also provides approximately 25 percent of the water used for irrigation globally (UNESCO, 2022; FAO AQUASTAT, 2023), helping to support food production for billions of people worldwide.

To put this figure in context: the vast majority of the world’s freshwater is locked in glaciers and ice caps. Liquid freshwater — water that is directly accessible for human use — represents a much smaller share of total global water. Groundwater accounts for virtually all of that accessible liquid fraction, making it by far the most significant freshwater resource available to people, agriculture, and ecosystems.

Despite this, groundwater is frequently overlooked in sustainability discussions. Its invisibility makes it difficult to monitor, easy to overexploit, and slow to attract policy attention. As populations grow, agricultural demand rises, and climate variability increases pressure on surface water sources, the case for understanding and managing groundwater becomes harder to ignore.

This article provides an accessible introduction to groundwater — what it is, why it matters, the threats it faces, and what can be done to protect it.

What Is Groundwater?

Groundwater is water that exists underground in saturated zones beneath the land surface, filling the pores and fractures in soils and rocks much the same way water fills a sponge (USGS, 2025). Over time, this water accumulates in underground formations called aquifers — geologic formations that contain sufficient saturated permeable material to yield significant quantities of water to wells and springs (USGS, 2025). The upper surface of the saturated zone is called the water table.

The process by which water enters the ground is known as infiltration. When rain falls or snow melts, some water flows across the surface as runoff into rivers and lakes, while the rest seeps downward through the soil and rock — a movement called percolation. As it descends, water passes through an unsaturated zone before reaching the water table. Below the water table, all pores, cracks, and spaces between rock particles are fully saturated with water; this is the saturated zone, and the water it contains is groundwater (USGS, 2025). Groundwater moves slowly, typically at rates of 7 to 60 centimetres per day within an aquifer, and as a result, water can remain stored underground for hundreds or even thousands of years.

Types of Aquifers

Unconfined Aquifers

Also known as water table aquifers, these are found near the surface and are directly recharged by infiltration from rainfall. They are more vulnerable to contamination from surface activities but are generally easier to access through shallow wells (USGS, 2025; UNESCO WWDR, 2022).

Confined Aquifers

These are deeper aquifer systems trapped between impermeable layers of rock or clay. Because the water is under natural pressure, it can rise on its own when a well is drilled — sometimes reaching the surface without any pumping at all. A well where this happens is called an artesian well: the pressure in the confined aquifer pushes water upward through the well shaft, in the same way that squeezing a sealed water bottle forces water out when you open the top. Artesian wells were historically important sources of water supply precisely because they required no mechanical pumping. Confined aquifers are generally less susceptible to surface contamination than unconfined ones, but they typically take much longer to recharge because they are sealed off from direct infiltration at the surface (USGS, 2025; UNESCO WWDR, 2022).

Fossil Aquifers

Some aquifers contain water that fell as rain or snow thousands or even millions of years ago, during periods when the climate was wetter than it is today. This water seeped underground and has been stored there ever since — long before any human civilisation existed. These are called fossil aquifers, or non-renewable aquifers. They work like a savings account that took thousands of years to fill: once the water is withdrawn, there is no rainfall cycle or recharge process that will refill them within any timeframe that matters to people alive today or in the foreseeable future. The Nubian Sandstone Aquifer System beneath Egypt, Libya, Sudan, and Chad is one of the best-known examples. It contains water that accumulated when the Sahara was green and wet — a climate that no longer exists. Managing fossil aquifers is therefore fundamentally different from managing renewable groundwater: every litre extracted is a litre permanently gone (UNESCO WWDR, 2022).

Why Groundwater Matters?

Groundwater does far more than supply drinking water. It supports food systems, industrial economies, and environmental health in ways that rarely receive public attention. Examining these functions helps explain why protecting and managing groundwater matters.

Groundwater and Drinking Water

Access to safe drinking water is one of the most fundamental requirements for human health and dignity. Groundwater plays a central role in meeting this need. According to the UNESCO World Water Development Report (2022), groundwater supplies around half of all water withdrawn for domestic use globally and is the primary source of drinking water for the majority of the world’s rural population.

In many lower-income countries and rural communities, piped water systems either do not exist or cannot be relied upon. In these contexts, wells and boreholes that tap groundwater are often the only available source of drinking water. The quality and availability of that groundwater directly determines the health outcomes of the communities that depend on it.

Groundwater and Food Security

According to the UNESCO World Water Development Report (2022) and FAO AQUASTAT (2023), groundwater supplies approximately 25 percent of the world’s irrigation water, helping farmers grow crops and sustain livestock, particularly during dry seasons and periods of limited rainfall.

Agriculture is the largest consumer of freshwater globally, and the reliability of food production in many regions depends on access to groundwater during dry periods when surface water sources are insufficient. In India, Pakistan, large parts of sub-Saharan Africa, and the Middle East, groundwater irrigation is not a supplement to surface water — it is the primary source. When groundwater declines, food production declines with it, with consequences that extend from household nutrition to national food security and global commodity markets.

Groundwater and Economic Development

Industries across multiple sectors depend on groundwater. Food processing, beverages, mining, textiles, pharmaceuticals, and construction all consume significant quantities of water in their operations. Schools, hospitals, hotels, and small enterprises in many parts of the world rely on private boreholes because public supply networks are unavailable or unreliable.

The economic consequences of groundwater insecurity can be severe. According to NITI Aayog’s Composite Water Management Index (2018), approximately 600 million people in India face high to extreme water stress, while groundwater levels continue to decline in several parts of the country. Separately, the World Bank projects that India could face a loss of approximately six percent of GDP by 2050 if its current path of water management does not change — a finding that points to the economic urgency of sustainable groundwater governance.

Groundwater and Ecosystem Health

Groundwater does not remain stationary underground. It moves slowly through aquifer systems and eventually discharges into rivers, lakes, springs, and wetlands. This process — known as baseflow — is what keeps rivers flowing during periods between rainfall events and through dry seasons. Many ecosystems depend entirely on this groundwater discharge to survive.

When groundwater is over-extracted, baseflow declines. Rivers shrink or dry up. Wetlands lose the water that sustains their biodiversity. Aquatic habitats disappear. The consequences extend not only to wildlife but to the human communities that depend on these ecosystems for water supply, food, and livelihood. Protecting groundwater is therefore also an act of ecosystem conservation.

What are the Threats to Groundwater?

Groundwater’s invisibility is both its strength and its vulnerability. Because it cannot be seen, the threats it faces often go unrecognised until the damage is already done. Knowing what these threats are is therefore a necessary starting point for any response.

Over-Extraction

The most widespread threat to groundwater is over-extraction — removing water from aquifers faster than natural processes can replenish them. As populations grow and demand for water rises, extraction often outpaces recharge rates. Water tables fall, wells must go deeper, and the long-term availability of groundwater as a resource comes under threat.

India offers one of the clearest illustrations of this challenge. According to NITI Aayog (2018), groundwater is being extracted faster than natural recharge rates across large parts of the country, particularly in northern agricultural states. The Chennai water crisis of 2019 — in which the city’s four main reservoirs nearly ran dry — showed what happens when surface water and groundwater reserves are depleted at the same time.

Contamination

Groundwater contamination occurs when pollutants enter aquifer systems from the surface or subsurface. Common sources include agricultural runoff carrying fertilisers and pesticides, industrial discharge of chemicals and heavy metals, leakage from waste disposal sites, and inadequate sanitation systems that allow human and animal waste to infiltrate the soil.

Contaminated groundwater is a direct risk to public health. According to NITI Aayog (2018), India is ranked 120th out of 122 countries on the water quality index, with approximately 70 percent of available water estimated to be contaminated. Once an aquifer is contaminated, remediation is technically complex, costly, and often only partially effective. Prevention is far more cost-effective than treatment.

Saltwater Intrusion

Coastal aquifers are groundwater systems that sit at the interface between land and sea. They are found along coastlines worldwide and supply drinking water and irrigation to millions of people in coastal communities. What makes them distinctive is their position: freshwater from inland recharge zones flows seaward through the aquifer, while saltwater from the ocean presses inward. Under natural conditions, the weight and pressure of the freshwater column holds the saltwater back, creating a stable boundary known as the freshwater-saltwater interface. This balance is delicate. Coastal aquifers are shallower, more accessible, and more exposed to surface activity than deeper inland systems — making them productive but also highly vulnerable to both contamination and salinisation.

When fresh groundwater is extracted faster than it is replenished, the boundary between freshwater and saltwater shifts inland. Seawater intrudes into the aquifer, rendering it unusable for drinking or irrigation — a process known as saltwater intrusion.

Once saltwater intrusion occurs, it is extremely difficult to reverse. Coastal communities in West Africa, South Asia, and the Pacific Islands are among those already experiencing this threat. As sea levels rise due to climate change and extraction pressure intensifies, saltwater intrusion is expected to become a more widespread problem.

Climate Change

Climate change affects groundwater in multiple and interconnected ways. Changes in rainfall patterns alter the timing and quantity of aquifer recharge. Longer and more intense droughts reduce infiltration and increase extraction pressure simultaneously. Flooding, paradoxically, can cause groundwater contamination when surface pollutants are driven underground by excessive rainfall.

Rising temperatures also increase evaporation, reducing the amount of precipitation that actually reaches aquifers. In some mountain regions, glacial retreat is removing a natural seasonal recharge mechanism that communities have relied upon for centuries. The combined effect is to make groundwater less predictable and, in many regions, less available — at the same time that demand is rising.

Global Examples of Groundwater Challenges

The challenges facing groundwater are not abstract. They are playing out in specific places, affecting real communities, and demanding practical responses. The following examples illustrate how groundwater insecurity can emerge across different contexts.

India

India faces one of the most acute groundwater crises in the world. According to NITI Aayog’s Composite Water Management Index (2018), approximately 600 million people face high to extreme water stress. Agricultural activity — which accounts for roughly 78 to 90 percent of India’s freshwater use depending on the region (FAO AQUASTAT) — has placed sustained pressure on aquifers across the country.

In several northern states, groundwater is being extracted faster than natural recharge rates, causing water tables to fall. Satellite data from NASA’s GRACE mission, cited in multiple peer-reviewed studies, has confirmed groundwater depletion in north-western India. The Chennai water crisis of 2019, in which the city’s four main reservoirs nearly ran dry, showed what urban water insecurity looks like in practice — and what happens when surface water and groundwater reserves decline together.

India’s situation carries lessons for the broader Global South, where similar pressures — population growth, agricultural intensification, uneven infrastructure, and climate variability — are changing the water security landscape.

North Africa

Beneath the deserts of Egypt, Libya, Sudan, and Chad lies one of the world’s largest known freshwater reserves — the Nubian Sandstone Aquifer System. Covering an area of approximately two million square kilometres, this fossil aquifer contains water that accumulated thousands of years ago when the Sahara had a much wetter climate.

Because the Nubian Aquifer is a fossil resource, its water is non-renewable on any human-relevant timescale. Extraction for agricultural and urban development projects in the region — most notably the Great Man-Made River project in Libya — has been drawing down reserves at rates that cannot be sustained. Managed under the International Shared Aquifer Resource Management (ISARM) framework coordinated by UNESCO and the IAEA (IAEA, 2021), the Nubian Aquifer shows both what groundwater can offer and what is lost when a finite resource is treated as though it has no limit.

Nigeria

Nigeria illustrates a different aspect of groundwater vulnerability — one centred on quality rather than quantity. According to UNICEF Nigeria (2021), over 60 million Nigerians lack access to safe water — one of the lowest safe water access rates in West Africa. This reflects not only infrastructure gaps but also groundwater quality risks created by unregulated extraction and poor sanitation.

In many Nigerian cities, private boreholes are the primary water source for households and businesses that cannot rely on public supply systems. Shallow boreholes are, however, vulnerable to contamination from pit latrines, open defecation, and surface runoff in densely populated areas. The 2022 Nigeria floods — which OCHA described as the worst since 2012, displacing over 1.4 million people (UNDP, 2023) — drove surface contaminants into groundwater systems and worsened the water quality crisis across multiple states.

South Africa

Between 2015 and 2018, Cape Town experienced one of the most severe droughts in its recorded history, extensively documented in peer-reviewed literature including Muller (2018) in Nature and by the World Bank (2018). As reservoir levels declined and concerns about “Day Zero” — the point at which municipal taps would be turned off — intensified, the city turned to groundwater as an emergency supplementary source.

The Table Mountain Group Aquifer, which underlies much of the Cape Peninsula, was identified as a potential buffer. Emergency borehole infrastructure was developed and activated. Conservation efforts ultimately helped the city avoid Day Zero, but the episode showed both what groundwater can offer in a crisis and how much is lost when it is treated as a last resort rather than part of long-term water planning.

Protecting Groundwater for Future Generations

Groundwater depletion and contamination are not inevitable. They result from management choices — or the absence of them. Protecting groundwater requires coordinated action across multiple levels of society, grounded in better data, clearer governance, and a clear understanding of groundwater’s value.

What Does Groundwater Security Mean for Citizens?

For individuals and communities, groundwater security directly affects the quality and reliability of drinking water, the productivity of local agriculture, and the health of the ecosystems that shape daily life. When groundwater is depleted or contaminated, the consequences are not shared equally — it is typically the poorest and most marginalised communities who lack the resources to find alternative sources and who bear the greatest burden.

Practical actions can make a meaningful contribution:

• Communities can practise rainwater harvesting to reduce reliance on borehole extraction and support natural aquifer recharge.

• Households and businesses should dispose of chemicals, pharmaceuticals, and waste through designated channels rather than drains or open ground.

• Citizens should report suspected illegal borehole drilling or industrial discharge to relevant environmental or water regulatory bodies.

• Reducing household water consumption through low-flow fittings and regular leak monitoring can make a meaningful difference.

• Individuals and communities can participate in local land stewardship initiatives that protect recharge zones such as wetlands and forested areas.

What Does Groundwater Security Mean for Institutions?

For governments, businesses, and development organisations, groundwater is both a resource to be used and a long-term liability if mismanaged. The scale of current groundwater challenges calls for coordinated institutional responses across all levels.

Institutions can act in the following ways:

• Integrated Water Resource Management (IWRM): Adopt cross-sectoral planning frameworks that manage surface water, groundwater, demand, and environmental flows together rather than in isolation.

• Groundwater governance: Establish and enforce regulatory frameworks that monitor extraction rates, protect recharge zones, and prevent contamination from agricultural and industrial sources.

• Climate-resilient infrastructure: Design water supply systems that account for groundwater variability under projected climate change scenarios, incorporating groundwater as a resilience buffer rather than an afterthought.

• Data transparency: Improve monitoring and public reporting on groundwater levels, quality, and extraction rates to enable evidence-based policy and informed public participation.

• Cross-border cooperation: Develop and strengthen agreements for the equitable management of shared aquifer systems, particularly in transboundary regions such as the Nubian Sandstone Aquifer.

Groundwater and the Sustainable Development Goals

Groundwater’s significance runs through many of the commitments in the 2030 Agenda for Sustainable Development. Protecting and managing it sustainably is not separate from achieving the Sustainable Development Goals, it is a condition for achieving several of them.

SDG 6 (Clean Water and Sanitation) places universal access to safe drinking water at the centre of the global development agenda. Groundwater — available in more places than any surface water source — cannot be excluded from efforts to meet this goal, particularly for rural populations and communities in the Global South that are not served by centralised infrastructure.

Beyond SDG 6, groundwater intersects with the following goals:

• SDG 2 (Zero Hunger): Approximately 25 percent of global irrigation water comes from groundwater. Sustainable aquifer management is essential to food security (UNESCO, 2022; FAO AQUASTAT, 2023).

• SDG 3 (Good Health and Well-being): Contaminated groundwater is a leading cause of waterborne disease. Safe groundwater directly reduces child mortality and supports community health outcomes.

• SDG 9 (Industry, Innovation and Infrastructure): Groundwater underpins manufacturing, mining, energy, and construction. Many enterprises in the Global South depend on private boreholes where public supply networks are absent or unreliable, making groundwater a foundation of industrial continuity.

• SDG 13 (Climate Action): Groundwater helps communities absorb the effects of droughts and extreme weather events. Climate change threatens recharge rates and aquifer stability, which means groundwater governance is also a climate adaptation concern.

• SDG 15 (Life on Land): Groundwater baseflow sustains rivers, wetlands, and terrestrial ecosystems. Over-extraction reduces biodiversity and disrupts ecological balance.

Conclusion

Groundwater is one of the most important resources on Earth and one of the least appreciated. Stored beneath the surface, it supplies drinking water to billions, supports the food systems that feed the world, sustains the ecosystems that regulate climate and biodiversity, and keeps the industrial and economic activities on which livelihoods depend.

Yet groundwater faces growing threats. Over-extraction is depleting aquifers faster than they can recharge. Contamination from agriculture, industry, and inadequate sanitation is rendering once-safe water sources unusable. Climate change is altering recharge patterns and, at the same time, driving up extraction pressure. And because groundwater is invisible, these trends often go unnoticed until the damage is already done.

The evidence in this article — from India’s declining water tables and Nigeria’s quality challenges to the non-renewable reserves of the Nubian Sandstone Aquifer and Cape Town’s emergency groundwater response — points to one conclusion: groundwater cannot continue to be treated as a resource without limits or as a backup supply to be drawn upon only in emergencies. It requires governance that anticipates problems, sustained investment in monitoring, and coordinated action across governments, businesses, communities, and individuals.

Groundwater is invisible, but its value is not. Understanding what it does, what threatens it, and what is required to protect it is a necessary step toward water security, climate resilience, and sustainable development for present and future generations.

References

Food and Agriculture Organization of the United Nations (FAO AQUASTAT). (2023). Global Water Data — Freshwater Withdrawals by Sector. https://www.fao.org/aquastat/en/

International Atomic Energy Agency (IAEA). (2021). Water Resources Management — Isotope Hydrology. https://www.iaea.org/topics/water-resources

Muller, M. (2018). Cape Town’s drought: Don’t blame climate change. Nature, 559, 174–176. https://doi.org/10.1038/d41586-018-05649-1

NITI Aayog. (2018). Composite Water Management Index. Government of India. https://social.niti.gov.in/uploads/sample/water_index_report2.pdf

United Nations Office for the Coordination of Humanitarian Affairs (OCHA). (2022). Nigeria — Floods Response: Flash Update. https://www.unocha.org/publications/report/nigeria/nigeria-floods-response-flash-update-1-last-updated-19-october-2022

UNICEF Nigeria. (2021). Water, Sanitation and Hygiene (WASH) — Nigeria Overview. https://www.unicef.org/nigeria/water-sanitation-and-hygiene

UNDP. (2023). Nigeria Flood Impact, Recovery and Mitigation Assessment Report 2022–2023. https://www.undp.org/nigeria/publications/nigeria-flood-impact-recovery-and-mitigation-assessment-report-2022-2023

UNESCO / UN-Water. (2022). United Nations World Water Development Report 2022: Groundwater — Making the Invisible Visible. UNESCO. https://www.unesco.org/reports/wwdr/2022/en

United States Geological Survey (USGS). (2025a). What is groundwater? USGS FAQs. https://www.usgs.gov/faqs/what-groundwater

United States Geological Survey (USGS). (2025b). Groundwater Basics. https://www.usgs.gov/mission-areas/water-resources/science/groundwater-basics

United States Geological Survey (USGS). (2025c). Groundwater Use — Aquifers and Withdrawals. https://www.usgs.gov/mission-areas/water-resources/science/groundwater-use

World Bank. (2018). Cape Town’s Day Zero crisis and urban water resilience. https://www.worldbank.org/en/topic/water

World Bank. (2018). Groundwater: Making the Invisible Visible — India Water Security Projections. Cited in NITI Aayog Composite Water Management Index (2018).

The passage of Hurricane Otis through Acapulco in 2023 marked a turning point in the discussion on water resilience in Mexico. Its rapid intensification overwhelmed any central planning model based on historical data, rendering entire drinking water, drainage, and sanitation systems inoperable in a matter of hours. The ensuing crisis was not only electrical or urban, but profoundly water-related: a lack of drinking water, the collapse of centralized treatment plants, and an immediate dependence on improvised, local solutions.

Otis demonstrated that, under extreme weather events, highly centralized water systems fail simultaneously, while recovery depends, de facto, on decentralized capacities: local catchment, community wells, modular treatment, and direct territorial management. This was not an anomaly, but an early sign of what happens when infrastructure is not designed for climate nonlinearity.

The current debate is often framed as a false dilemma: either the State completely controls water, or the door is opened to abusive privatization. This dichotomy impoverishes the discussion and blocks solutions. There is a third way, widely documented in institutional economics and increasingly urgent in the Mexican climate context: polycentric water governance.

Talking about polycentrism does not mean weakening the State, but rather redefining its role. Instead of being the sole operator, the State acts as an arbiter, guarantor of rights, rule designer, and evaluator of results. Daily management—collection, treatment, reuse, infiltration—is brought closer to the local level, where information is more precise and incentives clearer.

In a polycentric system, multiple actors make coordinated decisions: communities, municipalities, industrial users, cooperatives, local businesses, and public agencies. Each operates at its own scale, with defined responsibilities and verifiable metrics. The result is not chaos, but distributed resilience.

The Mexican water system was designed under a linear logic: extract, treat, use, drain. This architecture depends on large infrastructures, extensive networks, and a central authority that makes decisions remotely. Under normal conditions, it is already inefficient; under climate stress, it collapses.

The intense rains of 2025 clearly demonstrated this. When everything depends on collectors, pumping stations, and centralized treatment plants, a single bottleneck is enough to cause the system to fail. The water does not infiltrate, is not retained, is not reused: it becomes a risk.

What Does Polycentric Governance Mean for Water?

Applied to water, polycentric governance translates into very concrete principles:

• Decisions at the level closest to the impact: Those who experience the risk—floods, scarcity, pollution—must have decision-making power.

• Accountability linked to measurable results: It is not enough to comply with regulations; what matters is how much water is reused, how much infiltrates the ground, and how much is not extracted.

• Diversity of solutions depending on the territory: What works in the Valley of Mexico is not the same as what works in the Huasteca or the Bajío regions.

• Functional redundancy: If one node fails, others continue operating. That is resilience.

Concrete instruments for regenerative water management

Decentralization is not just rhetoric. It requires clear instruments:

• Legally recognized water microsystems: Rainwater harvesting, decentralized treatment, local reuse, constructed wetlands, and urban infiltration.

• Performance-based contracts

• Payment based on cubic meters reused, pollutant load removed, reduction in discharges, or network demand.

• Tariffs that recognize the value of local reuse

• If a community reduces pressure on the network, this should be reflected financially.

• Cooperatives and public-community partnerships

• Collective ownership, clear rules, transparency, and local maintenance.

• Management by micro-watershed, not just by administrative concession

Water follows the topography, not organizational charts.

These instruments align incentives: less extraction, more reuse; less discharge, more regeneration.

The circular economy applied to water is not about large “green” plants, but about closing cycles locally:

• Capture where it rains

• Treat where it is used

• Reuse where needed

• Infiltrate where the soil allows

This reduces losses, energy costs, and dependence on distant systems. Furthermore, it transforms wastewater into a resource and the land into an ally, not an enemy.

From this perspective, decentralized sanitation and closed-loop systems are not “alternative” solutions, they are critical infrastructure for the 21st century.

The State remains indispensable. But its role is changing:

It defines quality and monitoring standards.

It guarantees the human right to water.

It prevents abuses and the capture of the resource.

It facilitates information and coordination.

It rewards performance, not bureaucratic obedience. A strong State is not one that controls everything, but one that enables many to find effective solutions.

In conclusion, regenerating water requires freedom with responsibility. The Mexican water crisis will not be solved simply with more funding, more control, or more centralization. It will be solved when the institutional design allows water to be managed as what it truly is: a living, territorial, and dynamic system.

Regenerating water means accepting that there is no single solution, but rather many well-coordinated local solutions. It means trusting in the capacity of communities, municipalities, and users to care for what they use, provided there are clear rules and the right incentives.

After 2025, the message is clear: centralism has already shown its limitations.

Mexico’s water resilience depends on more decision-making centers, greater local responsibility, and more closed-loop systems.

Yes to the law. Yes to the state.

But also to freedom, polycentrism, and regeneration from the ground up.

Spanish

Del Centralismo al Poli centrismo: Cómo Regenerar El Agua en México Desde Abajo

México enfrenta una paradoja hídrica cada vez más evidente: mientras el discurso público insiste en el derecho humano al agua y en grandes planes nacionales, la realidad muestra sistemas frágiles, intermitentes y vulnerables a eventos extremos. Las inundaciones de 2025 en la Ciudad de México, el Estado de México y Veracruz, no fueron una anomalía, sino la manifestación de un modelo agotado: centralizado, rígido y desconectado del territorio.

El paso del huracán Otis por Acapulco en 2023 marcó un punto de inflexión en la discusión sobre resiliencia hídrica en México. Su rápida intensificación superó cualquier modelo de planeación central basado en datos históricos, dejando sin operación sistemas completos de agua potable, drenaje y saneamiento en cuestión de horas. La crisis posterior no fue solo eléctrica o urbana, sino profundamente hídrica: falta de agua potable, colapso de plantas centralizadas y dependencia inmediata de soluciones improvisadas y locales.

Otis demostró que, bajo eventos climáticos extremos, los sistemas hídricos altamente centralizados fallan de forma simultánea, mientras que la recuperación depende, de facto de capacidades descentralizadas: captación local, pozos comunitarios, tratamiento modular y gestión territorial directa. No se trató de una anomalía, sino de una señal temprana de lo que ocurre cuando la infraestructura no está diseñada para la no linealidad climática.

El debate actual suele plantearse como un falso dilema: o el Estado controla completamente el agua, o se abre la puerta a la privatización abusiva. Esta dicotomía empobrece la discusión y bloquea soluciones. Existe una tercera vía, ampliamente documentada en la economía institucional y cada vez más urgente en el contexto climático mexicano: la gobernanza policéntrica del agua.

Hablar de poli centrismo no significa debilitar al Estado, sino redefinir su rol. En lugar de ser operador único, el Estado actúa como árbitro, garante de derechos, diseñador de reglas y evaluador de resultados. La gestión cotidiana —captación, tratamiento, reúso, infiltración— se acerca al territorio, donde la información es más precisa y los incentivos más claros.

En un sistema policéntrico, múltiples actores toman decisiones coordinadas: comunidades, municipios, usuarios industriales, cooperativas, empresas locales y organismos públicos. Cada uno opera en su escala, con responsabilidades definidas y métricas verificables. El resultado no es caos, sino resiliencia distribuida.

El sistema hídrico mexicano fue diseñado bajo una lógica lineal: extraer, potabilizar, usar, drenar. Esta arquitectura depende de grandes infraestructuras, largas redes y una autoridad central que decide desde lejos. Bajo condiciones normales ya es ineficiente; bajo estrés climático, colapsa.

Las lluvias intensas de 2025 lo demostraron con claridad. Cuando todo depende de colectores, bombeos y plantas centralizadas, basta un cuello de botella para que el sistema falle. El agua no infiltra, no se retiene, no se reutiliza: se convierte en riesgo.

Qué significa gobernanza policéntrica aplicada al agua

Aplicada al agua, la gobernanza policéntrica se traduce en principios muy concretos:

Decisiones al nivel más cercano al impacto

Quien vive el riesgo, inundación, escasez, contaminación— debe tener capacidad de decisión.

Responsabilidad vinculada a resultados medibles

No basta con cumplir normas; importa cuánta agua se reutiliza, cuánta se infiltra, cuánta se deja de extraer.

Diversidad de soluciones según el territorio

Lo que funciona en el Valle de México no es lo mismo que en la Huasteca o el Bajío.

Redundancia funcional

Si falla un nodo, otros siguen operando. Eso es resiliencia.

Instrumentos concretos para una gestión regenerativa del agua

La descentralización no es retórica. Requiere instrumentos claros:

Microsistemas hídricos legalmente reconocidos

Captación pluvial, tratamiento descentralizado, reúso local, humedales construidos, infiltración urbana.

Contratos por desempeño

Pago por metros cúbicos reutilizados, carga contaminante removida, reducción de descargas o demanda a red.

Tarifas que reconozcan el valor del reúso local

Si una comunidad reduce presión sobre la red, debe reflejarse económicamente.

Cooperativas y asociaciones público-comunitarias

Propiedad colectiva, reglas claras, transparencia y mantenimiento local.

Gestión por microcuenca, no solo por concesión administrativa

El agua sigue la topografía, no los organigramas.

Estos instrumentos alinean incentivos: menos extracción, más reutilización; menos descarga, más regeneración.

La economía circular aplicada al agua no consiste en grandes plantas “verdes”, sino en cerrar ciclos localmente:

Captar donde llueve

Tratar donde se usa

Reutilizar donde se necesita

Infiltrar donde el suelo lo permite

Esto reduce pérdidas, costos energéticos y dependencia de sistemas lejanos. Además, convierte al agua de residuo en recurso y al territorio en aliado, no en enemigo.

Desde esta lógica, el saneamiento descentralizado y los sistemas de ciclo cerrado no son soluciones “alternativas”: son infraestructura crítica para el siglo XXI.

El Estado sigue siendo indispensable. Pero su función cambia:

Define estándares de calidad y monitoreo

Garantiza el derecho humano al agua

Evita abusos y captura del recurso

Facilita información y coordinación

Premia el desempeño, no la obediencia burocrática

Un Estado fuerte no es el que lo controla todo, sino el que permite que muchos resuelvan bien.

En conclusión, regenerar el agua exige libertad con responsabilidad. La crisis hídrica mexicana no se resolverá solo con más presupuesto, más control o más centralización. Se resolverá cuando el diseño institucional permita que el agua se gestione como lo que es: un sistema vivo, territorial y dinámico.

Regenerar el agua implica aceptar que no existe una solución única, sino muchas soluciones locales bien coordinadas. Implica confiar en la capacidad de comunidades, municipios y usuarios para cuidar lo que usan, siempre que existan reglas claras e incentivos correctos.

Después de 2025, el mensaje es claro: el centralismo ya mostró sus límites.

La resiliencia hídrica de México pasa por más centros de decisión, más responsabilidad local y más ciclos cerrados.

Ley sí. Estado sí.

Pero también libertad, poli centrismo y regeneración desde abajo.

This article was originally published in Spanish here.

A household installs a solar water heater once. A household makes food decisions multiple times each day, across decades, embedded in habit, family expectation, cultural meaning, and material constraint. Solar adoption responds to visible peer examples and to the alignment between technology and a recognized household problem. Dietary change requires altering routines that most households do not experience as decisions at all, but as the ordinary texture of daily life. Policy that treats these routines as considered individual choices tends to intervene at the wrong level, and the household remains unpersuaded.

Sustainable dietary behaviors for a green transition include adopting predominantly plant-based diets, reducing food waste across the household and supply chain, choosing seasonal and local foods where available, and attending to the energy and resource costs of food preparation. These practices reduce the environmental footprint of food systems and contribute to food security and nutritional adequacy for present and future generations. The relative importance of each dimension varies considerably by context, income level, and food system, a variation this piece will examine.

Treating Food as an Information Problem

The dominant policy approach to sustainable diets assumes that households, given sufficient information about the environmental costs of their food choices, will update their behavior accordingly. Where information is insufficient, price signals are expected to shift the calculation. Where both fall short, communication campaigns try to activate environmental concern.

Some of these tools have produced real results in specific contexts. But they consistently struggle to explain one of the most consistently observed problems in applied behavioral research: the intention-action gap. Surveys across multiple countries find that far more households report intentions to eat more sustainably than change what they eat. Sheeran’s (2002) meta-analysis of the intention-behavior relationship confirmed that forming an intention and acting on it are reliably distinct processes. The gap persists in dietary behavior because information is entering a system already structured by habit, social norms, and material conditions that operate largely below the level of deliberate choice.

Food as Practice, Not Choice

All the mechanisms that make dietary behavior resistant to policy share one underlying logic: existing food routines remain adaptive. They solve problems households face every day.

The first mechanism is habit. Most of us do not consciously decide what to eat each evening. We repeat routines triggered by familiar kitchens, shops, and schedules. Behavioral research on habit and behavior change, reviewed in depth by Verplanken and Wood (2006), established that these routines are activated by contextual cues rather than by deliberate intention. Interventions that target intention without disrupting the cues that trigger the routine will underperform, because the decision they are addressing is rarely consciously made at all.

The second mechanism is social and cultural embedding. Claude Fischler’s work on food and identity established that what a household eats is shaped by what counts as a proper meal in its cultural context, by what is served at celebrations, by what children will accept, and by the unspoken norms of the community around it. Research on the psychological justifications people offer for meat consumption has documented four recurring rationalizations: meat is natural, normal, necessary, and nice (Piazza et al., 2015). These are the active social work through which a food routine is maintained against challenges to its legitimacy. Environmental information about meat enters that system without necessarily disrupting it.

The third mechanism is time scarcity. Among the most consistent predictors of dietary patterns in rapidly urbanizing contexts is available time rather than environmental attitude. Working hours and commuting distances push households toward processed foods and convenience products requiring less preparation. Time scarcity also drives food waste: households that purchase more than they can prepare before items spoil are responding to procurement habits built around availability rather than planned consumption.

The household is also not a single decision-maker, and significant food consumption happens outside it entirely, in offices, schools, restaurants, and canteens. Who purchases food, who cooks, who decides the weekly menu, and who bears the labor cost of changing what is served are often different people with different constraints. Dietary change that requires more time or more planning imposes costs that fall unevenly, often on those who already carry a disproportionate share of domestic food labor.

China

The EAT-Lancet Commission’s planetary health diet (Willett et al., 2019), is one of the most influential sustainability-oriented dietary models in global policy discussions, functioning as a reference framework. Its framing centers on reducing animal-source food consumption in favor of plant-based alternatives. That framing travels well in contexts where per capita meat consumption is already high. It encounters a more complex behavioral reality in contexts where households are still moving toward the diets that the framework asks them to leave behind.

Across many countries undergoing economic development, rising incomes have historically been associated with dietary diversification and increased consumption of animal-source foods. Bennett’s Law, a long-observed pattern in nutritional economics, describes how dietary diversity rises with income, with animal-source foods increasing most rapidly in the middle-income range. China’s dietary transformation over the past four decades illustrates this at scale. Using longitudinal data from the China Health and Nutrition Survey, tracking household diets since 1989, Zhai and colleagues documented a sustained shift away from grain-based staples toward animal-source foods, fats, and processed products (Zhai et al., 2009; Zhai et al., 2014).

The mechanism is not income alone. Urban food environments, transformed by the expansion of markets, restaurant chains, and street food infrastructure, made animal-source foods far more available than they had been for rural households a generation earlier. Time scarcity in newly urbanized households accelerated the shift toward convenience foods. Social comparison reinforced meat-inclusive diets as a marker of improved welfare. The result is a dietary routine that becomes, within a generation, as habituated and as socially expected as the grain-based diet it replaced.

A policy that says “eat less meat” is addressed to a household for which eating more meat is a recent achievement and a legitimate aspiration. Environmental information about beef’s carbon footprint enters a routine in which beef also carries meaning about the household’s place in the world, and that meaning is not easily dislodged by a label or a campaign.

India

India has among the lowest per capita meat consumption figures of any large economy, according to FAO Food Balance Sheet data, and survey evidence from the National Family Health Survey (NFHS-5, 2019-21) confirms that a substantial share of the population follows predominantly plant-based diets for reasons of religion, culture, and economic constraint. The behavioral bottlenecks to more sustainable food systems in India are located elsewhere: in food loss across supply chains, in household food waste driven by storage constraints and procurement routines, in the energy costs of cooking, and in the food environment pressures of rapid urbanization.

The behavioral drivers of household food waste in Indian urban contexts include procurement habits built around fresh daily purchases rather than planned weekly shopping, storage infrastructure that cannot extend the life of perishable items, and social expectations around hospitality that favor preparing more food than will be consumed. These are behavioral and infrastructural challenges that EAT-Lancet framing was not designed to address.

The cooking dimension is also significant. The WHO estimates that approximately 2.3 billion people globally still rely on solid biomass fuels for cooking, and the IEA’s India Energy Outlook (2021) estimates that over 660 million people in India, roughly half the population, remain primarily dependent on such fuels. Health and environmental costs are substantial. The barriers to cleaner cooking fuel adoption involve trust in unfamiliar technologies, upfront cost, the sensory familiarity of traditional cooking methods, and the gendered distribution of cooking labor. These challenges require a framework built around the actual household rather than categories of analysis developed for a different food system.

What Behavioral Policy Can Realistically Change

If food behavior is produced through habits, social norms, and material infrastructures, the useful policy question becomes: where can we intervene in the systems that produce what people do?

The best evidence comes from institutional settings. School meal programs in several Scandinavian countries and in the United Kingdom have shifted default meal options from meat-inclusive to plant-based, with meat available as an opt-out. The results show measurable reductions in meat consumption with low levels of active resistance. The limitation matters: defaults work when an institution controls the food environment. The household dinner table has no institutional architect.

South Korea’s volume-based food waste fee system offers a different kind of evidence. Introduced in 1995, extended by a ban on food waste landfilling in 2005, and further strengthened by the nationwide rollout of a weight-based smart bin system from 2013, the system charges households according to the volume of food waste they generate. It works by redesigning the material and financial conditions within which the food routine takes place, making waste reduction the path of least resistance rather than the path requiring deliberate effort. Park and Lah (2015) analyzed the system’s trajectory across municipal solid waste data from 1986 to 2012, documenting sustained improvements in waste separation and recycling performance. The mechanism is behavioral through structural redesign: it targets the routine at the level where the routine is produced.

Where food consumption is partly symbolic, interventions that alter what the symbol communicates can reach further than those that add new information. The reported decline in shark fin soup consumption in China following WildAid’s social norm campaigns in the early 2010s illustrates this principle, though the concurrent government ban on shark fin at official functions makes causal attribution difficult.

Carbon labeling offers a more contested evidence base. Camilleri and colleagues (2019) found positive effects from carbon footprint labels in a US experimental context. Some subsequent real-world studies in restaurant and retail settings have produced more modest effects, and the evidence base remains contested. Carbon labeling may contribute at the margins as part of a broader food environment redesign, but it cannot carry the weight of a primary behavioral lever on its own.

What the Cases Suggest

In China, households adopted meat-inclusive diets because those diets solved real problems: status in a newly urbanized social environment, nutritional variety, the time pressures of working urban households. In India, households maintain fresh daily purchasing habits, generous hospitality portions, and biomass cooking systems because those routines solve the problems of limited refrigeration, social obligation, and familiar food preparation within specific material constraints. In South Korea, redesigning the conditions of the routine produced more durable change than attitudinal campaigns had.

The conceptual thread is this: households do not adopt diets. They maintain systems of everyday problem-solving. Every food routine solves something, whether that is time, cost, status, nutrition, hospitality, child acceptance, or social belonging. Changing dietary behavior means changing the solution to those problems, not merely changing a stated preference. This creates a direct bridge to solar installment: in both domains, households are maintaining or changing ways of living that solve the problems they face.

Implications for Citizens, Institutions, and Policy

For citizens: The intention-action gap in dietary behavior reflects the depth at which food routines are embedded in daily life, material systems, and cultural meanings. Approaching dietary change through the lens of context, what can be made easier, more affordable, or more normal, tends to produce more durable results than relying on motivation alone.

For institutions: Canteens, schools, hospitals, and public procurement systems are the settings where food environments can be directly redesigned. Behavioral research accumulated in settings where defaults and friction reduction have been tested suggests these tools outperform information campaigns when institutions control the food environment. Institutions that have this control have more leverage than they typically exercise.

For policymakers: Sustainable diet policy designed around reducing excess consumption needs significant adaptation before it can travel to contexts where the behavioral challenges differ. Food waste reduction, cooking infrastructure, supply chain loss, and nutritional transition each require frameworks built around the households and food systems that exist. The South Korean model suggests that pricing interventions targeted at daily routines, supported by appropriate infrastructure, can produce sustained change without relying on attitudinal change as the primary lever.

Closing Note

The recurring question of this series, what is the actor actually doing and why does the standard policy frame miss it, produces a layered answer in the case of dietary behavior. In China, the household is participating in a transition it experiences as progress. In India, it is managing a food system shaped by constraints the dominant framework was not designed to address. In South Korea, redesigning the conditions of the routine produced the most durable documented change.

Sustainable food policy succeeds not when it persuades people to care more, but when it redesigns the environments in which everyday decisions are made. That is where the leverage is, and where the next generation of behavioral work in this domain needs to go.

The next installment will examine household transport choices and the green transition, asking what the evidence shows about why public transport investments succeed in some cities and struggle in others even when economic conditions appear comparable.

References

Blackford, Becky. 2021. “Nudging Interventions on Sustainable Food Consumption: A Systematic Review”. The Journal of Population and Sustainability 5 (2):17–62. https://doi.org/10.3197/jps.2021.5.2.17.

Camilleri, A.R., Larrick, R.P., Hossain, S., and Patino-Echeverri, D. (2019). Consumers underestimate the emissions associated with food but are aided by labels. Nature Climate Change, 9, 53-58. https://doi.org/10.1038/s41558-018-0354-z

Fischler, C. (1988). Food, self and identity. Social Science Information, 27(2), 275-292. https://doi.org/10.1177/053901888027002005

International Institute for Population Sciences (IIPS). (2021). National Family Health Survey (NFHS-5), 2019-21: India. Mumbai: IIPS.

International Energy Agency (IEA). (2021). India Energy Outlook 2021. Paris: IEA. https://www.iea.org/reports/india-energy-outlook-2021

Park, S., and Lah, T.J. (2015). Analyzing the success of the volume-based waste fee system in South Korea. Waste Management, 43, 533-538. 10.1016/j.wasman.2015.06.011

Piazza, J., Ruby, M.B., Loughnan, S., Luong, M., Kulik, J., Watkins, H.M., and Seigerman, M. (2015). Rationalizing meat consumption: The 4Ns. Appetite, 91, 114-128. https://doi.org/10.1016/j.appet.2015.04.011

Sheeran, P. (2002). Intention—Behavior Relations: A Conceptual and Empirical Review. European Review of Social Psychology, 12(1), 1–36. https://doi.org/10.1080/14792772143000003.

Stoll-Kleemann, S., and Schmidt, U.J. (2017). Reducing meat consumption in developed and transition countries to counter climate change and biodiversity loss: A review of influence factors. Regional Environmental Change, 17(5), 1261-1277. https://doi.org/10.1007/s10113-016-1057-5

Verplanken, B., and Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy and Marketing, 25(1), 90-103. https://doi.org/10.1509/jppm.25.1.90

Willett, W., Rockström, J., Loken, B., et al. (2019). Food in the Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food systems. The Lancet, 393(10170), 447-492. https://doi.org/10.1016/S0140-6736(18)31788-4

Zhai, F., Wang, H., Du, S., He, Y., Wang, Z., Ge, K., & Popkin, B. M. (2009). Prospective study on nutrition transition in China. Nutrition reviews, 67(suppl_1), S56-S61. https://doi.org/10.1111/j.1753-4887.2009.00160.x

Zhai, F., Du, S., Wang, Z., Zhang, J., Du, W., and Popkin, B. (2014). Dynamics of the Chinese diet and the role of urbanicity, 1991-2011. Obesity Reviews, 15(S1), 16-26. https://doi.org/10.1111/obr.12124

Traditional power grids are not good enough for modern low-carbon cities. They only send energy one way, do not check energy use in real time. They also cannot work well with small clean energy (such as rooftop solar panels). These grids were made for large fossil fuel power stations. They struggle to balance clean energy supply, control busy-time energy use and fit smart cities’ needs. This inflexibility causes wasted energy, more emissions and blackouts, failing to help cities meet sustainability goals.

This article explains what a smart grid is and how it helps sustainable cities. It will answer several key questions. These include what a smart grid is, how it differs from traditional grids, and how it works in cities. It will also cover what stops it from being used, its benefits, and what people can do to use it.

What Is a Smart Grid?

A smart grid is a new and improved electricity system. It is safer, more efficient and greener than a traditional power grid. A smart grid is different from a conventional grid in many ways. It can use renewable energy such as wind and solar power better, which traditional grids cannot do well (Zhou et al., 2013). It also has fewer power cuts, so it is more reliable. Besides, it saves energy, cuts costs and produces less pollution.

In terms of management, traditional grids cannot fit modern development, but smart grids use smarter control methods (Zhou et al., 2013). Traditional grids only have one‑way power flow, while smart grids allow two‑way flow because of small power stations (Zhou et al., 2013). For load control, traditional grids use manual work, but smart grids use automatic control (Zhou et al., 2013). Smart grids also use new and accurate monitoring technology instead of old and slow systems (Zhou et al., 2013).

A smart grid mainly has three basic technologies: sensors, two‑way communication and automation. Sensors are very important because they collect real‑time data to find problems fast and protect the grid (Alonso et al., 2020, p. 1, p. 6). Two‑way communication helps different devices share information and send messages quickly when faults happen (Alonso et al., 2020, p. 3, p. 7). Automation can help the grid fix problems alone, isolate short circuits and reduce the need for human work (Alonso et al., 2020, p. 2, p. 3).

In urban sustainability, a smart grid follows several important core principles. It uses modern technology to provide safe electricity, save energy and reduce costs (Silva et al., 2025, p. 5). It only focuses on the energy field (Silva et al., 2025, p. 5). It helps cities cut CO₂ emissions by using renewable energy with ICT tools (Silva et al., 2025, p. 5). It allows flexible two‑way energy flow to balance supply and use (Silva et al., 2025, p. 6). It uses automatic systems to fix faults fast and reduce the influence of power cuts (Silva et al., 2025, p. 13). It also supports small energy resources like solar power, batteries and electric cars to make the grid more stable (Silva et al., 2025, p. 12).

Smart Grids for Solar and Wind

Smart grids solve many weak points of old power systems and let cities use more solar and wind energy on a large scale. Traditional power lines only send electricity one way from big power stations to homes and factories. This old design cannot work with many small solar panels and wind machines spread across urban areas (Kiasari et al., 2024, p.4).

Smart grids, however, allow electricity and data to move in two directions. This two-way system lets home solar equipment and small wind plants connect to the main power network. It also lets local residents join energy management work. This basic setup supports renewable energy whose power output often goes up and down (Kiasari et al., 2024, p.4). To balance changeable solar and wind power with people’s daily electricity use, Advanced Metering Infrastructure (AMI) acts as a key tool (Kiasari et al., p.4). AMI uses simple smart meters, communication wires and real-time data systems to set different electricity prices for different times. It encourages people to use power when solar or wind supply is enough, so the highest load pressure on grids becomes smaller.

AMI costs a lot to build. Rich cities with clear price rules can use it widely, but poor areas cannot afford this system easily (Kiasari et al., 2024, p.4). Even if residents change their power use habits, solar and wind power still changes all the time. Energy storage machines are therefore necessary parts of smart grids (Kiasari et al., 2024, p.8). Solar energy only appears in daytime and wind speed changes every hour, so their power supply is not stable. Smart grids use batteries, supercapacitors and pumped water storage to save extra power made at noon or on windy days. When sunlight or wind is weak, these storage devices send out saved electricity to meet people’s needs. Besides balancing supply and demand, storage equipment can quickly adjust voltage and frequency to keep the whole power system steady when more clean energy is used (Kiasari et al., 2024, p.15). Still, most storage machines cost much money and break down quickly, so people cannot use them together with renewable energy on a large scale.

AI Improves Energy Efficiency

AI tools fixed inside city smart grids have several easy ways to raise power efficiency and cut different kinds of wasted electricity (Veerraju et al., 2024, p.73). Old grid control cannot adjust power supply in real time correctly, which causes two main waste problems. The first problem is making too much power for possible high demand; the second is the bad match between solar, wind supply and citizens’ power needs. This research shows AI software solves these two issues with practical methods.

Machine learning tools such as ANN can guess future power use accurately, so less extra power is wasted. The study says ANN has 92% accuracy when it predicts city electricity demand (Veerraju et al., 2024, p.73). With correct forecasts, power workers only make the exact amount of electricity people need, instead of producing lots of spare power for unknown busy hours. This reduces wasted fossil fuel power and makes city power systems work better.

RL reinforcement learning saves the most energy among all tested AI tools. Test results show it can cut 32% of wasted city power and balance loads better (Veerraju et al., 2024, p.73). Unlike ANN and SVM which only finish one single job, RL improves power production, transport and home use at the same time. It balances changeable solar and wind power with normal electricity, so cities do not throw away clean energy when sun or wind is strong. The paper also states GA and SVM can help save power and find grid faults, which cuts total city energy waste further (Veerraju et al., 2024, p.73).

AI helps home and shop users save power from the demand side. Smart meters and sensors collect lots of real-time data for AI systems, so city people can check and change their own power use easily (Veerraju et al., 2024, p.74). This citizen participation pushes people to run large electric machines when solar or wind power is enough, not during busy peak hours. It makes power curves less steep and lowers the need for extra emergency power, so less energy is wasted for sudden high demand.

AI acts as a common simple way to deal with unstable solar and wind output in cities (Veerraju et al., 2024, p.74). Solar and wind power levels change every day. Old grids either throw away extra clean power or start low-efficiency backup power plants to fill gaps, both causing huge waste. AI prediction and optimization tools can guess how much green power will be produced and send power around properly. In this way, cities make full use of solar and wind resources and waste less clean energy.

AI for Grid Stability

AI control systems inside smart grids solve three main power troubles: balanced load, peak power use and blackout risks (Fangzong & Nishter, 2024, p.2). This system uses LSTM and GRU prediction tools together with simple control rules. Old power networks only act after faults happen, but new AI systems plan power use early to match supply and demand well and stop grid breakdown risks. For balanced load across the whole city grid, the study uses two easy steps: real-time power guess and automatic power sharing rules (Fangzong & Nishter, 2024, p.9). LSTM and GRU can guess future power use correctly. The system then sends power evenly to different power lines and makes good use of small solar and wind machines. Old grids often carry too much load on some lines because they cannot predict power use early. This AI method spreads power equally in city grids and avoids heavy load on single areas.

To manage peak power demand, the system uses two common ways: energy storage and user demand response (Fangzong & Nishter, 2024, p.12). Storage saves extra power when people use little electricity and releases power during busy hours. The system also offers small rewards to ask people to shift power use to quiet time. Test data shows this method cuts peak load by 10% on average; in July, peak power drops from 160 MW to 140 MW (Fangzong & Nishter, 2024, p.16). Fewer peak hours mean power companies do not need to open extra old power plants, which saves energy and money for cities.

When talking about blackout prevention, the paper names two main reasons for power cuts: big voltage changes or wrong low guesses of power demand (Fangzong & Nishter, pp.12,19). The AI system watches voltage all day and changes solar/wind power output to limit voltage changes from 4%–7.5% to 3%–5% (Fangzong & Nishter, 2024, p.16). Besides, correct forecasts stop workers from preparing too little power. If prediction numbers are lower than real use, power shortages and blackouts will appear. The mix of prediction and real-time watch builds safety barriers and stops small faults turning into city-wide power cuts.

Real Cases of Urban Smart Grids

Real energy projects in different countries give clear proof that smart microgrids (a main kind of smart grids) help build sustainable cities. This study talks about both successful and failed city cases, which show the technology’s strong points and real limits (Almihat & Munda, 2025, p.1). These examples cover North America, Europe and Asia, and each one tells us how smart grids cut carbon gas, balance power and protect cities during bad weather.

San Diego, Barcelona and Seoul are three big cities that widely use smart microgrids (Almihat & Munda, pp.7–8). These cities build local solar and wind systems plus machines that make heat and power together. They do not need to buy power from faraway big power plants. This way cuts energy lost during long-distance transport and lowers city carbon gas. This local energy supply fits low-carbon targets of sustainable cities and proves smart grids work well in large urban areas.

Brooklyn Microgrid in New York is a good community example (Almihat & Munda, p.1, p.15). It uses simple blockchain tools to let local people sell extra solar power to each other without big power firms. This easy trade encourages more people to use clean energy and build self-sufficient green neighbourhoods. Besides, microgrids in busy New York areas can cut peak power use by 15% (Almihat & Munda, pp.7–8). Cities do not need to spend lots of money to fix old power lines, so smart grids save city costs while reaching green goals. For disaster safety, microgrids in New Jersey kept working when the main grid broke during Hurricane Sandy in 2012 (Almihat & Munda, pp.7–8). They supplied power to hospitals and public services, so cities become safer during disasters. Two Asian real cases also show this good point: Sendai microgrid in Japan still worked in the 2011 earthquake and tsunami, and Jeju smart grid project in South Korea tests large automatic clean energy systems for island cities (Almihat & Munda, p.15). A German village called Wildpoldsried also runs well; its wind and solar systems make extra green power to share with other places (Almihat & Munda, p.15).

However, London’s energy plan is a negative real example (Almihat & Munda, p.1). Its smart microgrid projects move very slowly because of strict energy rules and not enough public money. This case shows smart grid technology alone cannot build sustainable cities. Governments must make suitable rules and give enough money support at the same time. All these global city examples clearly show smart grids have practical value for sustainable cities, but their results depend on local policies and financial support.

Recommendations for Stakeholders

Local and national governments should use three types of tools together to spread smart grid technology quickly (Almihat & Munda, 2025, p.1). They can offer PPP projects, low-interest loans and tax cuts to lower investment risks for small local power projects. Long-term tools such as carbon trading and green bonds can attract investors who want stable long-term profits (Almihat & Munda, pp.19-20). Governments also need to shorten official waiting time for small community energy plans and make the same power rules across all cities. Different local trading rules stop small solar sales from growing fast, while shared grid safety rules help people copy good microgrid designs and cut building costs (Almihat & Munda, pp.5-6, p.20). Constant tax breaks for green projects are necessary for companies to balance profit and environmental goals (Negi et al., 2025, p.12).

Green investors should buy and hold certified green bonds for a long time, instead of looking for fast short-term cash (Negi et al., pp.14-15). All money from these bonds can only go to clean energy and low-carbon buildings. Outside checks stop “greenwashing”, where firms pretend to be eco-friendly without real pollution cuts (Negi et al., p.2, p.11). Investors should choose businesses with long-term green plans, not firms that only care about quick gains.

Factory owners should join demand response groups via virtual power plants instead of working alone (Wang et al., 2025, p.798). Factories use 66% of total city electricity and can reduce power use for short periods without losing much output. Steel and cement plants can share load-cut tasks together. Group work makes more adjustable power and lowers each factory’s daily costs, so businesses are happy to join grid support plans for many years (Wang et al., p.799).

Transport and power planners must design electric bus infrastructure and power lines at the same time (Yetkin et al., 2024, p.2425). Separate designs create big charging peaks and break grid balance. Using bus batteries to store extra green power is cheaper than changing power station output. Governments and city councils should carry out long-term joint investment: add more electric buses, upgrade batteries and widen power cables to avoid future grid overload.

Ending Note

Technically, AI systems reduce peak load by around 10% (Fangzong & Nishter, 24). Economically, PPP and green bonds cut construction costs and create stable profits (Negi et al., 2025). Environmentally, cities use more clean energy and release less greenhouse gas. Socially, community microgrids strengthen disaster resistance, though high costs and inconsistent local rules act as barriers seen in London’s slow projects (Almihat & Munda, 2025). To gain full benefits, governments need unified standards and tax discounts; investors choose certified green assets; industries cooperate through virtual power plants (Wang et al., 2025).

References

Almihat, M. G. M., & Munda, J. L. (2025). The Role of Smart Grid Technologies in Urban and Sustainable Energy Planning. Energies, 18(7), 1618. https://doi.org/10.3390/en18071618

Alonso, Mónica & Amaris, Hortensia & Alcala-Gonzalez, Daniel & R., Diana. (2020). Smart Sensors for Smart Grid Reliability. Sensors. 20. 2187. 10.3390/s20082187. (PDF) Smart Sensors for Smart Grid Reliability

Fangzong, W., & Nishter, Z. (2024). Innovative Load Forecasting Models and Intelligent Control Strategy for Enhancing Distributed Load Levelling Techniques in Resilient Smart Grids. Electronics, 13(17), 3552. https://doi.org/10.3390/electronics13173552

Kiasari, M., Ghaffari, M., & Aly, H. H. (2024). A Comprehensive Review of the Current Status of Smart Grid Technologies for Renewable Energies Integration and Future Trends: The Role of Machine Learning and Energy Storage Systems. Energies, 17(16), 4128. https://doi.org/10.3390/en17164128

Long Wang, Tingzhe Pan, Yi Wang, Xin Jin, Heyang Yu, Wangzhang Cao. A virtual power plant scheduling strategy considering cooperative participation of multiple industrial users in demand response[J]. AIMS Energy, 2025, 13(4): 798-818. doi: 10.3934/energy.2025029 A virtual power plant scheduling strategy considering cooperative participation of multiple industrial users in demand response

Negi, P., Jaiswal, A. & Rekunenko, I. Green bonds and corporate value creation: a review of financial, market, and environmental performance. Discov Sustain 6, 1106 (2025). https://doi.org/10.1007/s43621-025-01834-1

Silva, N. S. e., Castro, R., & Ferrão, P. (2025). Smart Grids in the Context of Smart Cities: A Literature Review and Gap Analysis. Energies, 18(5), 1186. https://doi.org/10.3390/en18051186

Veerraju, Dr & Giridhar, Munigoti & SatyavedaSomepalli, & Reddy, Dr & Reddy, Dr.P. & Reddy, Dr.P.Nagasekhar. (2024). AI-powered Smart Grids: Energy Optimization. Nanotechnology Perceptions. 73-84. 10.62441/nano-ntp.vi.3613. (PDF) AI-powered Smart Grids: Energy Optimization

Yetkin, M., Augustino, B., Lamadrid, A.J. et al. Co-optimizing the smart grid and electric public transit bus system. Optim Eng 25, 2425–2472 (2024). https://doi.org/10.1007/s11081-023-09878-w

Zhou, J., He, L., Li, C., Cao, Y., Liu, X. & Geng, Y. (2013). What’s the difference between traditional power grid and smart grid? — From dispatching perspective. 2013 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC), Hong Kong, China, pp. 1–6. https://doi.org/10.1109/APPEEC.2013.6837107

ESG fragmentation Across Regions

ESG fragmentation refers to the growing disconnect, inconsistency, and lack of a universal framework in how Environmental, Social, and Governance (ESG) standards, laws, and data are defined and applied globally. Instead of a unified approach, companies now navigate a complex, overlapping web of localized rules, compliance costs, and reporting metrics. Regional divergence is the clearest driver of ESG fragmentation. In the European Union, the CSRD and its accompanying European Sustainability Reporting Standards (ESRS) remain the most detailed and prescriptive disclosure regime in the world, built on a double-materiality basis that requires companies to report both how sustainability issues affect their finances and how their operations affect people and the environment (Crowell & Moring LLP, 2026). However, the regime has itself been in flux: the EU’s Omnibus I Directive, finalized in February 2026, sharply narrowed CSRD’s scope to companies with more than 1,000 employees and €450 million in net turnover, and the European Commission adopted a simplified version of the ESRS on 3 July 2026 that cuts mandatory datapoints substantially while retaining the double-materiality principle (European Commission, 2026; Latham & Watkins, 2026).

The United States presents an almost opposite trajectory. Following the change in federal posture toward climate disclosure, the SEC ended its legal defence of its Climate Disclosure Rule, leaving climate reporting to a patchwork of state-level laws viz. California’s SB 253 and SB 261, chief among them. However, these state level laws themselves face ongoing litigation over their validity and timing (A&O Shearman, 2026). The result is a landscape where a company’s disclosure obligations can change materially depending on which US state it does business in, layered on top of whatever international frameworks it also follows (ESG Dive, 2026). Figure 1 attempts to identify contemporary challenges related to global divergence in ESG implementation

Beyond the EU and US, momentum has shifted toward the ISSB’s IFRS S1 and S2 standards as a global baseline. Adoption has accelerated quickly: by July 2025, sixteen jurisdictions had adopted the ISSB standards on a mandatory or voluntary basis; by January 2026 that had risen

to twenty-one; and by April 2026, twenty-eight jurisdictions representing roughly 60% of global GDP had done so, with a further twelve in active adoption processes (S&P Global Sustainable1, 2026). Markets including Japan, Brazil, Nigeria, Turkey, Malaysia, and Hong Kong have each adopted the standards with local modifications to timing and scope (ESGsource, 2026).

Types of ESG Reporting Systems

Part of the complexity multinational companies face comes from the sheer variety of system types. These systems must navigate simultaneously, each with a different legal status and purpose.

• Voluntary frameworks, such as the GRI Standards and CDP, are non-binding but widely adopted to build stakeholder credibility and comparability on impact-focused topics.

• Market-led initiatives, historically led by the Task Force on Climate-related Financial Disclosures (TCFD) and now absorbed into the ISSB, are driven by investor demand for financially material climate information rather than legal mandate.

• Mandatory national regulations, such as the EU’s CSRD/ESRS regime or California’s emissions-disclosure statutes, are legally binding and enforceable, typically carrying audit and penalty requirements.

• Hybrid systems, such as the UK’s evolving Sustainability Disclosure Requirements, blend voluntary adoption with a gradual transition toward regulatory enforcement.

• Finally, global convergence platforms, principally the ISSB standards, together with alignment work between the ISSB and GRI, and IOSCO’s endorsement of the ISSB baseline across its 130-plus member jurisdictions aim explicitly to reduce duplication by giving companies a single reference point that multiple regulators can point to (Teil Firms, 2026).

A company operating across two or three of these system types at once, which is now the norm for most multinationals, must reconcile not just different metrics, but different underlying logics of why disclosure is required in the first place.

Core Barriers to ESG Comparability

ESG comparability focuses towards the commonality and differences among the guidelines and legal mandates. Sometimes even if companies are willing to disclose extensively, several structural barriers make their disclosures difficult to compare across frameworks and jurisdictions. The table (Table 1) below summarizes the most significant of these barriers.

Thus, varied materiality approaches and inconsistent data quality tend to create the largest practical headaches. This appears because ESRS requires double materiality while ISSB applies a financial-materiality lens, a single data point. For instance, biodiversity impact in a supply chain agenda that may behave as reportable parameter under one regime and immaterial under another, even though the underlying facts have not changed (KEY ESG, n.d.). Meanwhile, the absence of standardized digital tagging means that even where two companies disclose the same figure, a human analyst, rather than software, is often still required to locate and reconcile it across PDF reports.

Cost of Fragmentation

ESG Fragmentation is not merely an inconvenience but it carries a measurable operational cost. Sustainability and finance teams increasingly report that they already hold the underlying data viz. climate figures, governance minutes, supplier due-diligence records, workforce metrics. Still, it must manually rebuild that same evidence into different formats for each of CSRD, ESRS, GRI, ISSB, SASB, TCFD, and an expanding set of customer and lender questionnaires each reporting cycle (Teil Firms, 2026). That duplication consumes resources that could otherwise go toward improving data quality or pursuing substantive sustainability performance, and it lengthens and complicates external assurance, since auditors must verify similar facts presented in materially different structures (GIS User, 2026).

Fragmentation also elevates greenwashing risk. Where disclosure requirements diverge, companies face a genuine risk of presenting the same underlying performance in ways that appear inconsistent or even contradictory across markets. They, not necessarily through, intent to mislead, but project so because reporting definitions differ. Regulators in multiple jurisdictions have already brought enforcement actions on this basis, including a widely discussed 2026 fine by Luxembourg’s financial regulator concerning a fund’s sustainability strategy disclosures (Proskauer Rose LLP, 2026). The reputational and legal stakes of getting cross-framework consistency wrong are rising in parallel with the number of frameworks in force.

ISSB, EFRAG, and the Search for a Global Baseline

Recognizing these costs, international standard-setters have been working toward interoperability rather than a single unified framework, which remains politically unlikely in the near term. The most significant milestone remains the creation of the ISSB itself, which absorbed the TCFD and the Sustainability Accounting Standards Board (SASB) into a single investor-focused baseline (S&P Global Sustainable1, 2026). Ongoing collaboration between the ISSB and the European Financial Reporting Advisory Group (EFRAG), which develops the ESRS, is intended to ensure that EU and global disclosures can substantially map onto one another — ESRS’s climate standard (E1) already aligns closely with IFRS S2, even though the EU’s double-materiality basis remains broader than the ISSB’s financial-materiality lens (Latham & Watkins, 2026).

Technology is expected to play a growing role in bridging what policy convergence cannot fully resolve on its own. Structured digital tagging (XBRL) and “collect once, report many times” data architectures allow companies to maintain a single internal dataset and generate framework-specific outputs from it, reducing the manual rebuilding that currently drives much of the cost of fragmentation (Teil Firms, 2026). The diagram (Fig.3) below summarizes this trajectory at a high level.

None of this suggests fragmentation will disappear soon. The EU’s own Omnibus process shows that even a single jurisdiction’s rules can shift twice within an 18-month window, and the US remains an outlier relative to the ISSB’s growing global footprint (ESGsource, 2026). But the direction of travel is paving ahead to more jurisdictions converging around a common baseline, and more technical work aimed at interoperability rather than duplication — gives companies a reasonably clear target to build toward.

Scope and Solution

This study is limited to corporate sustainability disclosure frameworks and regulations as they stood as of mid-2026, with particular focus on the EU (CSRD/ESRS), the United States (federal and state climate rules), and the ISSB’s global baseline. It does not cover sector-specific voluntary certifications (e.g., B Corp) or supply-chain-only frameworks, and jurisdictional detail is necessarily illustrative rather than exhaustive given how quickly individual rules are still changing.

However, organizations facing variable disclosure guidelines are best served by a three-part response.

• First, build a single internal sustainability data architecture — a “collect once” model — mapped against the specific datapoints required by each framework the company is actually subject to, rather than maintaining parallel, framework-specific data collection processes.

• Second, prioritize adoption of the ISSB standards as the organizing baseline wherever legally permissible, since a growing majority of relevant jurisdictions are converging around it, and layer jurisdiction-specific requirements (such as ESRS double materiality or California’s emissions rules) on top of that baseline rather than starting from scratch for each.

• Third, invest early in digital tagging and assurance-ready documentation, since both regulatory scrutiny and the operational cost of manual reconciliation are rising together; the earlier this infrastructure is built, the lower the marginal cost of responding to the next round of regulatory change.

References

A&O Shearman. (2026). US ESG trends: Fragmentation, backlash and energy security. https://www.aoshearman.com/en/insights/sustainability-outlook-2026/esg-trends-in-the-us-navigating-fragmentation-backlash-and-energy-security

Crowell & Moring LLP. (2026, March 10). EU sustainability reporting revamp: Key updates to the CSRD and the CS3D from the Omnibus I Directive. https://www.crowell.com/en/insights/client-alerts/eu-sustainability-reporting-revamp-key-updates-to-the-csrd-and-the-cs3d-from-the-omnibus-i-directive

ESG Dive. (2026, January 29). Companies face fragmented climate risk disclosure landscape in 2026. https://www.esgdive.com/news/corporate-climate-risk-disclosure-landscape-2026-challenges-data-outlook/810854/

ESGsource. (2026). ISSB adoption tracker 2026: Which countries require IFRS S1 & S2. https://esgsource.com/page/issb-adoption-tracker-2026

European Commission. (2026, July 3). Commission adopts revised sustainability reporting standards to reduce administrative burdens for EU businesses while maintaining high-quality disclosures. https://finance.ec.europa.eu/news/commission-adopts-revised-sustainability-reporting-standards-reduce-administrative-burdens-eu-2026-07-03_en

GIS User. (2026, April 8). How modern ESG reporting tools are closing the gap between data and disclosure. https://gisuser.com/2026/04/how-modern-esg-reporting-tools-are-closing-the-gap-between-data-and-disclosure/amp/

KEY ESG. (n.d.). A complete guide to ESG compliance and regulation. https://www.keyesg.com/article/esg-compliance

Latham & Watkins. (2026). EU sustainability: State of play — the conclusion of the Sustainability Omnibus process. https://www.lw.com/en/insights/eu-sustainability-state-of-play-the-conclusion-of-the-sustainability-omnibus-process

Proskauer Rose LLP. (2026, April 2). Risk #4: Five guidelines for ESG disclosures in 2026 — choose your words carefully. https://www.proskauer.com/blog/risk-4-five-guidelines-for-esg-disclosures-in-2026-choose-your-words-carefully

S&P Global Sustainable1. (2026, May 13). May 2026 — Where does the world stand on ISSB adoption? https://www.spglobal.com/sustainable1/en/insights/research-reports/issb-q2-2026

Teil Firms. (2026, May 13). ESG reporting convergence: ISSB, TCFD & global compliance strategy. https://www.teilfirms.com/blog/esg-reporting-is-converging-around-issb-and-tcfdbut-most-companies-are-still-rebuilding-the-same-report-repeatedly

Tourism is one of the world’s largest and fastest-growing industries, bringing people, cultures, and places closer together. Every year, millions of people travel for leisure, business, education, and adventure. It helps in creating jobs, support local businesses, and boost economies. For many countries, especially those that depend heavily on tourism, the industry is a major source of income and economic development. Despite its economic benefits, tourism has increasingly placed pressure on natural ecosystems, local communities, and cultural heritage through unsustainable practices. As a result, there is a growing need to manage tourism in a way that promotes economic development while safeguarding environmental integrity, cultural identity, and social well-being. According to the United Nations Tourism (UN Tourism), when managed responsibly, tourism can promote economic growth, encourage cultural exchange, and contribute to sustainable development.

Sustainable tourism is a development approach that seeks to ensure tourism activities generate long-term economic, environmental, and social benefits while minimizing adverse impacts on destinations and host communities. As tourism continues to expand globally, its contribution to economic development has become increasingly significant. However, unmanaged tourism can also result in environmental degradation, cultural commodification, biodiversity loss, and unequal economic distribution. Consequently, sustainable tourism has emerged as a comprehensive framework for promoting responsible tourism development that balances the interests of visitors, local communities, businesses, and the natural environment (United Nations Environment Program (UNEP) and United Nations World Tourism Organization (UNWTO, 2005).

The United Nations Tourism (UN Tourism) defines sustainable tourism as “tourism that takes full account of its current and future economic, social and environmental impacts, addressing the needs of visitors, the industry, the environment and host communities” (UN Tourism, as cited by the United Nations Sustainable Development Goals Knowledge Platform). This definition emphasizes that sustainability should not be viewed as a separate form of tourism but rather as a guiding principle applicable to all tourism activities, regardless of their scale or destination. Sustainable tourism therefore requires an integrated approach that balances economic prosperity, environmental stewardship, and social equity to ensure the long-term viability of tourism destinations.

In simple terms, sustainable tourism means traveling and managing tourism in a way that protects the environment, respects local cultures and communities, and supports the local economy without compromising the ability of future generations to enjoy the same destinations. It encourages responsible tourism practices that create long-term benefits for both people and the planet.

Tourism is one of the world’s fastest-growing industries and plays a significant role in creating jobs, generating income, and supporting economic development. According to the World Travel & Tourism Council, the travel and tourism sector contributed approximately 10% of global Gross Domestic Product (GDP) and supported around one in every ten jobs worldwide in 2023 (WTTC, 2024). These figures demonstrate the importance of tourism as a driver of economic growth. However, the rapid expansion of tourism has also increased pressure on the environment, local communities, and cultural heritage.

One of the major concerns is climate change. Tourism activities, particularly air travel, road transport, and accommodation, consume large amounts of energy and produce greenhouse gas emissions. The tourism sector is estimated to account for approximately 8–10% of global greenhouse gas emissions, making it an important contributor to climate change (Lenzen et al., 2018). As global tourism continues to grow, reducing the sector’s carbon footprint has become increasingly important.

Another challenge is plastic pollution and waste generation. Popular tourist destinations often experience large volumes of plastic waste, food waste, and other forms of pollution, especially during peak travel seasons. If waste is not managed properly, it can damage ecosystems, pollute oceans and rivers, and threaten wildlife (United Nations Environment Programme (UNEP, 2021).

Tourism can also lead to overtourism, where too many visitors place excessive pressure on a destination. Overcrowding can increase traffic congestion, raise living costs for local residents, damage historical sites, and reduce the overall quality of life within host communities. In addition, excessive tourism may contribute to the loss of biodiversity through habitat destruction, overuse of natural resources, and disturbance of wildlife (UN Tourism, n.d.).

Sustainable tourism provides a practical solution to these challenges. It encourages governments, businesses, local communities, and travelers to adopt responsible practices that protect natural resources, preserve cultural heritage, support local businesses, and create long-term economic opportunities. By balancing economic development with environmental conservation and social well-being, sustainable tourism helps ensure that destinations remain attractive and resilient for future generations (UNEP and UNWTO, 2005).

Sustainable tourism is based on three interconnected pillars: environmental sustainability, social sustainability, and economic sustainability. These pillars provide a comprehensive framework for ensuring that tourism contributes to long-term development while minimizing its negative impacts on destinations and local communities. Rather than focusing solely on economic growth, sustainable tourism seeks to create a balance between protecting the environment, improving people’s well-being, and generating lasting economic benefits (UNEP & UNWTO, 2005). Recent studies also emphasize that achieving sustainability in tourism requires integrating these three dimensions into tourism planning, policy, and management.

Environmental Sustainability

Environmental sustainability focuses on protecting natural resources and reducing the environmental impacts of tourism. Healthy ecosystems—including forests, mountains, wetlands, oceans, and wildlife habitats are essential for many tourism destinations. Conserving these ecosystems helps preserve biodiversity and ensures that natural attractions remain available for future generations (UN Tourism, n.d.).

Sustainable tourism encourages the protection of wildlife by promoting responsible tourism activities that avoid disturbing animals and their natural habitats. It also emphasizes reducing waste through recycling, proper waste management, and minimizing the use of single-use plastics. Tourism businesses are increasingly encouraged to adopt renewable energy sources, such as solar and wind energy, to reduce greenhouse gas emissions. Likewise, water conservation measures, including efficient water use, rainwater harvesting, and wastewater treatment, help protect limited freshwater resources, particularly in destinations facing water scarcity (GSTC, 2024; UNEP & UNWTO, 2005). Recent research further highlights that environmental sustainability remains the foundation of sustainable tourism because the quality of natural resources directly influences destination competitiveness and visitor satisfaction.

Social Sustainability

Social sustainability aims to improve the quality of life of local communities while preserving their cultural identity and traditions. Tourism should foster mutual respect between visitors and host communities rather than disrupting local lifestyles or cultural values (UNEP & UNWTO, 2005).

An important principle of social sustainability is community participation, where local residents actively contribute to tourism planning and decision-making. This helps ensure that tourism development reflects community needs and creates shared benefits. Sustainable tourism also promotes fair employment by encouraging safe working conditions, equal opportunities, and fair wages for tourism workers. Furthermore, respecting the rights of Indigenous peoples and protecting cultural heritage including historical sites, traditional knowledge, languages, festivals, and local customs helps preserve the unique identity of destinations while strengthening cultural pride (GSTC, 2024). Contemporary tourism research suggests that greater community involvement leads to more resilient destinations and improves both resident well-being and visitor experiences.

Economic Sustainability

Economic sustainability ensures that tourism generates long-term economic benefits that are shared fairly among local communities and businesses. Instead of creating short-term profits for only a few stakeholders, sustainable tourism encourages travelers to support locally owned hotels, restaurants, handicraft businesses, farms, and tour operators. This allows more tourism income to remain within the local economy and creates stronger economic linkages (UN Tourism, n.d.).

The economic pillar also promotes decent work by creating stable employment opportunities, fair wages, and opportunities for skills development. Sustainable tourism contributes to poverty reduction, particularly in rural and marginalized communities where tourism can become an important source of income. In addition, it supports inclusive economic growth by ensuring that women, youth, Indigenous communities, and small enterprises have equal opportunities to participate in and benefit from tourism development. According to recent literature, economically sustainable tourism strengthens local resilience, improves livelihoods, and supports broader sustainable development objectives by creating inclusive and diversified local economies.

Why Sustainable Tourism Matters More Than Ever

In recent years, sustainable tourism has become more important than ever due to growing environmental, social, and economic challenges. The rapid expansion of tourism, combined with climate change, increasing resource consumption, and changing traveler expectations, has highlighted the need for a more responsible approach to tourism development. While tourism continues to create employment and support economic growth, its long-term success depends on protecting the natural and cultural resources on which it relies (UNEP & UNWTO, 2005; García-García et al., 2023).

Climate Change and Environmental Challenges

Climate change is one of the greatest threats facing the tourism industry today. Rising global temperatures, changing weather patterns, wildfires, floods, sea-level rise, and the melting of glaciers are already affecting many popular tourist destinations. Coastal areas are experiencing beach erosion, while coral reefs are suffering from coral bleaching due to warming oceans. These environmental changes not only threaten biodiversity but also reduce the attractiveness of destinations that depend on nature-based tourism (Intergovernmental Panel on Climate Change (IPCC, 2023). As tourism both contributes to and is affected by climate change, adopting sustainable practices has become essential for reducing emissions and improving the resilience of destinations.

Addressing Overtourism and Protecting Destinations

Another growing concern is overtourism, where excessive numbers of visitors place pressure on local infrastructure, ecosystems, and communities. Popular destinations such as Venice, Bali, and Mount Everest have experienced overcrowding, increased waste generation, traffic congestion, and damage to natural and cultural heritage. Overtourism can reduce residents’ quality of life and negatively affect visitors’ experiences. Sustainable tourism addresses these issues by encouraging responsible destination management, visitor control measures, and conservation efforts that protect destinations while maintaining their long-term attractiveness (UN Tourism, n.d.).

Changing Expectations of Travelers

Travelers are also becoming more aware of the environmental and social impacts of their travel choices. Many tourists now seek authentic cultural experiences, environmentally responsible accommodations, and businesses that support local communities. This shift has encouraged tourism operators to adopt sustainable practices, improve resource efficiency, reduce waste, and obtain sustainability certifications. As sustainability increasingly influences travel decisions, responsible tourism is becoming both an ethical responsibility and a competitive advantage for destinations and businesses (World Travel & Tourism Council (WTTC, 2024).

A Shared Responsibility

Achieving sustainable tourism requires the collective efforts of governments, tourism businesses, local communities, and travelers. Governments play an important role by developing policies, protecting natural and cultural heritage, and investing in sustainable infrastructure. Tourism businesses can reduce their environmental footprint through energy efficiency, waste reduction, and responsible sourcing. Local communities should actively participate in tourism planning to ensure that development reflects their needs and values. At the same time, travelers can contribute by respecting local cultures, supporting local businesses, minimizing waste, and making environmentally responsible travel choices. Through shared responsibility and collaboration, sustainable tourism can continue to generate economic opportunities while protecting destinations for future generations (GSTC, 2024; UNEP & UNWTO, 2005).

Real-Life Examples of Sustainable Tourism

Sustainable tourism is not just a concept it is already being successfully practiced in many parts of the world. Countries such as Bhutan and Costa Rica show that tourism can support economic growth while protecting nature and preserving local culture.

Bhutan

Bhutan follows a “High Value, Low Volume” tourism policy, which focuses on attracting responsible travelers instead of large numbers of visitors. Tourists pay a Sustainable Development Fee (SDF), which helps fund environmental conservation, education, healthcare, and cultural preservation. This approach has allowed Bhutan to protect its unique heritage while ensuring that tourism benefits both the country and its people (Rinzin et al., 2007; Royal Government of Bhutan, 2022).

Costa Rica

Costa Rica has become a global leader in ecotourism by protecting large areas of its forests and wildlife. Through its Certification for Sustainable Tourism (CST) program, the country encourages tourism businesses to adopt environmentally friendly practices while supporting local communities. This has helped Costa Rica grow its tourism industry without compromising its rich biodiversity (Honey, 2008; ICT, 2024).

Challenges of Sustainable Tourism

While sustainable tourism offers many benefits, putting it into practice is not always easy. One of the biggest challenges is finding the right balance between economic growth, environmental protection, and the well-being of local communities. In many places, tourism development focuses heavily on increasing visitor numbers and short-term income, sometimes at the expense of natural resources and community interests (UNEP & UNWTO, 2005).

Another major issue is over tourism. When too many visitors arrive at the same destination, local infrastructure can become overwhelmed. Traffic increases, waste accumulates, ecosystems suffer, and residents may feel that their quality of life is declining. Popular destinations around the world have struggled with overcrowding and the pressure it places on both nature and local culture (UN Tourism, n.d.).

Climate change also presents a serious challenge. Tourism contributes to greenhouse gas emissions through transportation, accommodation, and other travel-related activities. At the same time, tourism destinations themselves are increasingly affected by rising temperatures, sea-level rise, floods, and extreme weather events (UNEP & UNWTO, 2005).

In addition, many small tourism businesses may not have the financial resources or technical knowledge needed to adopt sustainable practices. Some travelers are also unaware of how their choices affect local communities and the environment. These challenges show that achieving sustainable tourism requires cooperation, awareness, and strong long-term planning.

Opportunities in Sustainable Tourism

Despite these challenges, sustainable tourism creates valuable opportunities for destinations, businesses, and communities. As travelers become more environmentally and socially conscious, many are looking for authentic experiences that support local culture and protect nature. This growing demand encourages tourism businesses to adopt more responsible practices and improve their sustainability efforts (GSTC, 2024).

Sustainable tourism can also help strengthen local economies. When visitors stay in locally owned hotels, eat at local restaurants, and buy products made by local artisans, more of the tourism income remains within the community. This creates jobs, supports small businesses, and improves local livelihoods (UNEP & UNWTO, 2005).

Another important opportunity is innovation. Investments in renewable energy, waste reduction systems, water conservation, and digital tourism management can make destinations more efficient and resilient. Sustainable tourism can also generate funding for wildlife conservation, heritage preservation, and community development projects.

Countries such as Costa Rica and Bhutan have shown that tourism can support economic development while protecting natural and cultural heritage. Their experiences demonstrate that sustainability is not only an environmental responsibility but also a smart long-term strategy for tourism growth.

Recommendations For Sustainable Tourism Development

Building a more sustainable tourism industry requires action from governments, tourism organizations, businesses, local communities, and travelers. Each group has an important role to play.

For Policy Makers

Governments should strengthen sustainable tourism policies, invest in eco-friendly infrastructure, encourage renewable energy use, and improve public transportation systems. They should also monitor tourism impacts regularly and provide incentives for businesses that adopt sustainable practices (UNEP & UNWTO, 2005).

For Local Tourism Bodies

Tourism authorities should work closely with local communities when planning tourism development. Promoting lesser-known destinations can help reduce overcrowding, while awareness campaigns can encourage visitors to travel responsibly and respect local culture and the environment.

For Travelers

Travelers can make a significant difference by reducing waste, conserving water and energy, respecting local customs, and supporting locally owned businesses. Choosing accommodations and tour operators that follow sustainability standards is another practical way to travel more responsibly (GSTC, 2024).

For Tourism and Hospitality Businesses

Hotels, tour operators, restaurants, and other tourism businesses should focus on reducing single-use plastics, improving energy and water efficiency, sourcing products locally, and managing waste responsibly. Providing sustainability training for employees and involving local communities in tourism activities can further strengthen positive impacts (GSTC, 2024).

Conclusion

Tourism has the power to bring people together, create jobs, and support local economies, but its long-term success depends on how responsibly it is developed and managed. As this article has shown, sustainable tourism is about finding the right balance between economic growth, environmental protection, and the well-being of local communities. By protecting natural resources, preserving cultural heritage, and encouraging responsible travel, we can ensure that tourism continues to benefit both people and the planet.

The future of tourism is a shared responsibility. Governments, businesses, local communities, and travelers all have an important role to play in making tourism more sustainable. Whether it is through better policies, responsible business practices, or mindful travel choices, every effort contributes to protecting destinations for future generations. Sustainable tourism is not about traveling less, it is about traveling better, so that the places we enjoy today remain vibrant, authentic, and welcoming for years to come.

References

García-García, A. R., Jacobo-Hernández, C. A., Ochoa-Jiménez, S., & Valdez-del Río, S. (2023). Sustainable development and tourism: A review of the literature in WoS from 2001 to 2020. Sustainability, 15(24), 16805. https://doi.org/10.3390/su152416805

Global Sustainable Tourism Council. (2024). GSTC Destination Criteria. https://www.gstc.org/gstc-criteria/

Honey, M. (2008). Ecotourism and sustainable development: Who owns paradise? (2nd ed.). Island Press.

Instituto Costarricense de Turismo. (2024). Certification for Sustainable Tourism (CST). https://www.ict.go.cr/en/sustainability/certification-for-sustainable-tourism.html

Intergovernmental Panel on Climate Change. (2023). Climate Change 2023: Synthesis Report. IPCC. https://www.ipcc.ch/report/ar6/syr/

Lenzen, M., Sun, Y. Y., Faturay, F., Ting, Y. P., Geschke, A., & Malik, A. (2018). The carbon footprint of global tourism. Nature Climate Change, 8(6), 522–528. https://doi.org/10.1038/s41558-018-0141-x

Ministry of Finance, Royal Government of Bhutan. (2022). Sustainable Development Fee Act of Bhutan, 2022.

Rinzin, C., Vermeulen, W. J. V., & Glasbergen, P. (2007). Ecotourism as a mechanism for sustainable development: The case of Bhutan. Environmental Sciences, 4(2), 109–125. https://doi.org/10.1080/15693430701365420

United Nations. (n.d.). Sustainable tourism. Sustainable Development Goals Knowledge Platform. https://sdgs.un.org/topics/sustainable-tourism

United Nations Environment Programme. (2021). From pollution to solution: A global assessment of marine litter and plastic pollution. https://www.unep.org/resources/pollution-solution-global-assessment-marine-litter-and-plastic-pollution

United Nations Environment Programme, & United Nations World Tourism Organization. (2005). Making tourism more sustainable: A guide for policy makers. United Nations Environment Programme. https://www.unep.org/resources/report/making-tourism-more-sustainable-guide-policy-makers

United Nations Tourism. (n.d.). Sustainable development. https://www.unwto.org/sustainable-development

World Travel & Tourism Council. (2024). Economic Impact Research 2024. https://wttc.org/research/economic-impact

Systems thinking is the ability to understand relationships, patterns, feedback loops, trade-offs, and long-term consequences. UNESCO describes systems thinking competency as the ability to recognise and understand relationships, analyse complex systems, understand how systems are embedded across different domains and scales, and deal with uncertainty (UNESCO, 2017). This definition is especially relevant to sustainability because sustainability challenges rarely sit inside one subject, one department, one country, or one generation.

This is why systems thinking has become very important in business and technology. The World Economic Forum identifies systems thinking as a continuing core skill for the future of work, alongside analytical thinking, technological literacy, resilience, creativity, leadership, and lifelong learning (World Economic Forum, 2025). Harvard Business Review has also argued that innovation can create unintended social and environmental costs when leaders fail to examine the wider system in which decisions operate (Bansal and Birkinshaw, 2025).

The same message appears in classic management thinking. Peter Senge described systems thinking as the fifth discipline of the learning organisation because it helps people see patterns and interrelationships rather than isolated events (Senge, 2006). Donella Meadows also explained that the behaviour of a system often comes from its structure, including feedback loops, delays, rules, information flows, and incentives (Meadows, 2008). In simple terms, systems thinking helps us move from asking, “What happened?” to asking, “Why does this keep happening?”

This matters deeply for sustainability education. A student may learn that electric vehicles reduce tailpipe emissions. That is true, but a systems thinker also asks where battery minerals come from, how mining affects communities and ecosystems, how electricity is generated, how batteries are reused or recycled, and whether transport demand itself needs to be redesigned. A student may learn that recycling plastic is good. A systems thinker asks why so much plastic is produced in the first place, why products are designed for short life cycles, how packaging systems are governed, and who pays the cost when waste leaks into rivers and oceans.

Without systems thinking, sustainability education can become a list of good actions. Plant trees. Recycle. Use less energy. Buy eco-friendly products. These actions may help, but they do not automatically change the system that created the problem. Systems thinking pushes learners to examine causes, incentives, behaviours, power structures, design choices, and unintended consequences.

This is also important because technology alone cannot solve sustainability challenges. Artificial intelligence, data analytics, automation, clean energy, smart logistics, and carbon accounting platforms can support sustainability, but they must be used responsibly within the right system. If AI improves efficiency but increases energy demand, there is a trade-off. If automation reduces waste but removes jobs without reskilling workers, the solution may not be socially sustainable. If a company reduces emissions in its own operations but shifts the burden to suppliers, the problem has not disappeared. It has moved to another part of the system.

Recent research on sustainability in computing education supports this point. A systematic literature review found that sustainability in computing education is still developing and that many programmes need stronger engagement with the systemic nature of sustainability challenges (Peters et al., 2023). This is important because digital systems now shape transport, finance, manufacturing, education, communication, energy management, agriculture, and consumer behaviour. Future technologists therefore need to understand how to build efficient systems, as well as how those systems affect people and the planet.

Systems thinking also helps learners understand that sustainability has social, environmental, and economic dimensions. A cheap product may carry a hidden labour cost. A convenient delivery service may carry a carbon cost. A fast fashion item may carry a water, waste, and worker safety cost. A profitable business decision may carry a future regulatory or reputational cost. Sustainability education must help learners see these hidden costs rather than judging success only by price, speed, or short-term convenience.

For business leaders, systems thinking improves decision-making. MIT Sloan Executive Education notes that decisions in complex organisations can create ripple effects across supply chains, incentives, decision rights, information flows, and operating assumptions (MIT Sloan Executive Education, 2026). This is why sustainability cannot be left only to a sustainability department. It is connected to procurement, finance, product design, operations, marketing, human resources, logistics, risk management, governance, and customer behaviour.

For educators, systems thinking changes how sustainability should be taught. Instead of teaching sustainability only as a separate environmental topic, it should be embedded into business, science, technology, economics, geography, design, ethics, civic education, and leadership. Students should be asked to map supply chains, trace product life cycles, identify stakeholders, analyse trade-offs, examine local community problems, and design solutions that consider both immediate and long-term impacts.

A classroom project on food waste, for example, in addition to measuring how much food is thrown away should also examine farming practices, supermarket standards, pricing, packaging, consumer behaviour, school lunch systems, composting, hunger, logistics, and policy. Likewise, a lesson on clothing should discuss recycling or donation, as well as material choices, labour conditions, overproduction, marketing pressure, consumer culture, durability, repair, resale, and end-of-life management.

Systems thinking also helps reduce greenwashing. When learners understand systems, they are less likely to accept simple claims such as “eco-friendly”, “carbon neutral”, “natural”, or “sustainable” without evidence. They learn to ask: What boundary was used? What emissions were counted? What happened upstream and downstream? Were social impacts considered? Was biodiversity affected? Is this a real solution, or is the impact being shifted somewhere else?

Most importantly, systems thinking gives learners agency. Many young people and professionals feel overwhelmed by climate change, biodiversity loss, inequality, waste, and rapid technological change. Systems thinking does not make these problems smaller, but it makes them more understandable. It shows that problems are produced by systems, and systems can be redesigned.

The future needs people who can connect the dots. It needs engineers who understand ethics, business leaders who understand climate risk, teachers who connect sustainability to everyday life, policymakers who consider long-term consequences, and consumers who understand hidden impacts. It needs professionals who can work across disciplines and ask better questions before rushing into solutions.

Sustainability education must therefore move beyond awareness. Awareness tells us that a problem exists. Systems thinking helps us understand why the problem exists, how it is connected to other problems, and where responsible action can create lasting change.

Therefore, the most valuable learners and leaders will not be those who can see relationships, understand consequences, question assumptions, and act with responsibility. That is why systems thinking must be at the heart of sustainability education.

References

Bansal, T. and Birkinshaw, J. (2025) ‘Why You Need Systems Thinking Now’, Harvard Business Review, September. Available at: https://hbr.org/2025/09/why-you-need-systems-thinking-now (Accessed: 8 July 2026).

Meadows, D.H. (2008) Thinking in Systems: A Primer. White River Junction, VT: Chelsea Green Publishing.

MIT Sloan Executive Education (2026) ‘What is Systems Thinking in Business?’, MIT Sloan Executive Education Blog, 11 May. Available at: https://executive.mit.edu/blog/what-is-systems-thinking-in-business.html (Accessed: 8 July 2026).

Peters, A.-K., Capilla, R., Coroama, V.C., Heldal, R., Lago, P., Leifler, O., Moreira, A., Fernandes, J.P., Penzenstadler, B., Porras, J. and Venters, C.C. (2023) ‘Sustainability in Computing Education: A Systematic Literature Review’, arXiv:2305.10369. Available at: https://arxiv.org/abs/2305.10369 (Accessed: 8 July 2026).

Senge, P.M. (2006) The Fifth Discipline: The Art and Practice of the Learning Organization. Revised edn. New York: Doubleday/Currency.

UNESCO (2017) Education for Sustainable Development Goals: Learning Objectives. Paris: UNESCO. Available at: https://www.unesco.org/en/articles/education-sustainable-development-goals-learning-objectives (Accessed: 8 July 2026).

World Economic Forum (2025) The Future of Jobs Report 2025. Geneva: World Economic Forum. Available at: https://www.weforum.org/publications/the-future-of-jobs-report-2025/in-full/3-skills-outlook/ (Accessed: 8 July 2026).