The intersection of artificial intelligence (AI) and environmental sustainability has become a critical topic as organizations adopt AI at scale. ISO 26000 – the international standard for social responsibility – provides a useful framework to examine this issue. In particular, ISO 26000’s guidance on “The Environment” outlines how organizations should address their environmental impacts. This includes principles of environmental stewardship such as pollution prevention, sustainable resource management, energy efficiency, waste reduction, and greenhouse gas reduction. In essence, companies deploying AI are encouraged to minimize their environmental footprint, echoing the same responsibilities expected in any other aspect of operations.
Applying these principles to AI means that businesses and governments need to align AI development with sustainability goals. This involves assessing the carbon emissions, energy usage, and resource consumption associated with AI systems and taking active steps to mitigate negative impacts. The concept of “sustainable AI” (or “Green AI”) has emerged in recent years to encapsulate this idea. Green AI refers to developing and using AI in a manner that maximizes efficiency and innovation while minimizing environmental harm. It calls for a balance where AI’s benefits to society and the economy are achieved alongside a commitment to environmental responsibility.
Crucially, environmental responsibility in AI is now seen as part of broader ESG (Environmental, Social, and Governance) goals for organizations. Just as ISO 26000 links social responsibility to areas like labor practices and human rights, it also links to environmental care – ensuring that AI initiatives do not occur in a vacuum but are part of a company’s holistic commitment to sustainable development. In practice, this means companies might conduct environmental impact assessments for AI projects, invest in cleaner technologies, and report AI-related emissions in sustainability reports. By following ISO 26000’s environmental guidance, AI innovators can help ensure that technological progress does not come at the cost of ecological well-being.
(This article is part of the AI and CSR Series, entry #5, building on prior discussions of AI in labor practices, human rights, and governance.)
AI technologies – especially machine learning and large-scale data analysis – require substantial computing power. This translates into significant energy consumption and a notable carbon footprint. For example, training a single state-of-the-art AI model can consume enormous amounts of electricity. Recent research estimated that training OpenAI’s GPT-3 (a large language model) used about 1,287 MWh of electricity, emitting roughly 502 metric tons of CO₂ – equivalent to the annual emissions of over 100 cars. And that figure only covers the training phase. Once deployed, AI models continue to draw power for inference (i.e. making predictions or answering queries), which can account for up to 60% of an AI system’s total energy usage. In practical terms, everyday uses of AI add up: for instance, a single ChatGPT query has been found to use 100× more energy than a typical Google search. Multiply these demands across millions of users and thousands of AI applications, and the energy and carbon costs mount quickly.
Beyond carbon emissions, AI’s infrastructure also consumes other resources. Data centers that house servers for AI and cloud computing not only use electricity but also vast amounts of water for cooling. On average, a data center consumes about 1.7 liters of water per kWh of energy used. This means AI growth can put pressure on water supplies, especially in regions where water is scarce. Additionally, manufacturing the hardware for AI – chips, servers, and cooling systems – involves material extraction and e-waste considerations. All these factors contribute to AI’s overall environmental impact.
In Asia, these challenges are particularly pronounced. The Asia-Pacific region is experiencing a digital boom, with surging demand for cloud services, data storage, and AI-driven applications. This has led to a rapid expansion of data centers across the region. Southeast Asia, for example, is now a major data centre hub – but the growth comes with a “dark side” of high power usage and emissions. In Singapore, data centres currently consume about 7% of the nation’s total electricity and produce an estimated 82% of the tech sector’s carbon emissions. These striking figures prompted the Singapore government to temporarily halt new data centre construction in 2019 and rethink its approach to digital infrastructure growth. It’s a vivid illustration that AI’s digital backbone has a hefty environmental price tag, especially if the energy powering it comes from fossil fuels. In many Asian countries, electricity grids still rely heavily on coal and natural gas, meaning the carbon intensity of running AI tends to be higher than in regions with cleaner grids. Moreover, in tropical climates, data centres require even more electricity for cooling to maintain safe operating temperatureseco-business.com. All told, without interventions, AI’s carbon footprint and resource demands could undermine sustainability efforts. This recognition is driving interest in “greening” the tech ecosystem to make AI part of the solution rather than part of the problem.
To address the environmental impact of AI, stakeholders across Asia are turning their attention to the infrastructure that powers AI – namely, data centres and cloud computing facilities. Green data centres have become a focal point for policy and innovation. A green data centre is one designed for maximum energy efficiency and minimum environmental impact, often by using advanced cooling techniques, efficient hardware, and renewable energy sources. In the context of AI, greener data centres directly translate to a smaller footprint for AI services, since these facilities provide the computing power for AI algorithms.
Singapore offers a leading example of pushing towards sustainable AI infrastructure. After grappling with the strain of data centre energy use on its climate goals, Singapore has enacted measures to ensure future digital growth is aligned with sustainability. Under the nationwide Green Plan 2030 (which charts Singapore’s sustainability targets) and an ambition to achieve net-zero emissions around mid-century, authorities recognized that AI and digitalization must advance in an eco-friendly way. In 2022, Singapore lifted its earlier moratorium on new data centers but introduced stringent environmental standards for any new facilities. Notably, the country launched the world’s first sustainability standard for data centres operating in tropical climates. This standard – developed by government agencies like IMDA along with industry partners – sets benchmarks for energy efficiency, renewable energy adoption, and operational resilience in data centre design. The goal is to ensure that even in a hot, humid climate, data centres (and by extension, the AI systems they support) can run with dramatically lower electricity and water usage. Complementing this, Singapore’s Green Data Centre Roadmap provides a strategic blueprint to guide operators in reducing environmental impact while meeting the needs of a growing digital economy.
These efforts are already influencing data centre investments and operations across Asia. Technology companies in the region are also stepping up. For instance, Google has invested in multiple data centres in Singapore, incorporating features like efficient cooling and lower water consumption in its newest facilities. Major cloud providers are increasingly committing to power Asian data centers with renewable energy where possible, or to offset their emissions. This is part of a broader industry trend: companies are integrating AI projects into their corporate sustainability agendas. Google reports that its data centres today are 3× more energy-efficient than just five years ago, and that new AI training optimizations have made some models’ emissions 1,000× lower than before. Similarly, these companies are aiming for ambitious targets such as 24/7 carbon-free energy by 2030 for all their operations, which would make the continuous power usage of AI essentially carbon-neutral.
Elsewhere in Asia, countries like Malaysia, Indonesia, and India are exploring ways to attract digital investment while managing the energy burden. Malaysia, for example, is emerging as a data centre hub due to lower energy costs, but there is growing emphasis that this growth must be accompanied by a shift to renewables to cut emissions. In Japan and South Korea, tech firms are investing in green building designs and even locating some data centers in cooler climates or underground to naturally reduce cooling needs. The rise of green data centres in Asia represents a concrete step toward sustainable AI: by cleaning up the back-end infrastructure, the AI services delivered to businesses and consumers can be made much more environmentally friendly. It’s a clear case of how ISO 26000’s call for sustainable resource use and energy efficiency can be translated into practice through engineering and policy solutions.
While we work to reduce AI’s own environmental footprint, it’s equally important to recognize how AI can be a powerful tool to drive environmental sustainability. The same computational prowess that makes AI resource-intensive can be harnessed to solve complex environmental problems – a field often referred to as climate tech when digital technology is applied to climate and ecological challenges. Guided by a responsibility framework like ISO 26000, organizations are looking not only to green AI itself but also to leverage AI in service of the environment.
One promising area is using AI to optimize energy use and reduce emissions across various sectors. For example, AI algorithms can analyze and manage power grids, balancing supply and demand in real time to integrate more renewable energy. In fact, a joint study by Google and BCG in 2023 suggested that scaling up proven AI solutions globally could help achieve about 20% of the emissions reductions needed by 2030 to meet climate targets. These gains come from improvements such as better data monitoring of emissions, AI-assisted integration of renewables into power grids, and smarter systems that cut waste. A concrete case is traffic management: Google has piloted an AI project that synchronises traffic lights to reduce idling time for vehicles, thereby saving fuel and cutting urban carbon emissions – trials are underway in Indonesia and India and have shown promising results in reducing unnecessary red-light stops.
AI is also being deployed in climate adaptation and resilience efforts across Asia. In South Asia, where extreme weather events are intensifying, researchers and non-profits are using AI for early warning systems. For instance, in India, the Gujarat Mahila Housing Sewa Trust (a non-profit) is using AI to predict flood risks in local communities, enabling thousands of at-risk residents (especially women farmers) to prepare and avoid hazards. Such predictive analytics showcase AI’s ability to process vast climate datasets – from satellite imagery to sensor networks – and identify patterns that humans might miss, potentially saving lives and livelihoods.
Another domain of climate tech is precision agriculture, where AI helps farmers in Asia optimize crop planting and irrigation with climate forecasts, thus improving yield while conserving water. In Indonesia and Malaysia, AI-driven analytics are aiding in forest conservation, analyzing satellite photos to detect illegal deforestation or predict fire outbreaks in peatlands. Likewise, Singapore has explored AI for urban sustainability – using machine learning models to design energy-efficient buildings and smart cooling systems in the tropics. These examples align perfectly with ISO 26000’s encouragement for organizations to “support environmental initiatives and sustainable development” as part of their social responsibility. By investing in AI solutions that directly address environmental challenges (from climate change mitigation to biodiversity monitoring), companies and governments can offset some of AI’s costs with significant societal benefits.
It’s worth noting that AI is not a silver bullet – these climate tech applications require cross-sector collaboration, community engagement, and sound policy to be effective. But they illustrate a hopeful synergy: when guided responsibly, AI and sustainability can reinforce each other. By innovating for good, AI can help predict environmental risks, optimize resource use, and even discover new solutions (like materials for carbon capture or more efficient renewable energy systems). Such positive uses of AI underscore why outright restriction of AI is not the answer; instead, ethical frameworks like ISO 26000 help maximize the upside while curbing the downside.
As awareness grows, environmental criteria are becoming integral to how we evaluate AI and tech initiatives. Investors, regulators, and the public are increasingly asking hard questions about the environmental impact of technology. This is where ESG (Environmental, Social, Governance) considerations come in. Organizations are now expected to align AI development with their ESG commitments, particularly the Environmental component. In practice, this means companies might measure and disclose the carbon footprint of their AI operations, set targets to reduce it, and link executive accountability or investment decisions to hitting those targets.
Frameworks like ISO 26000 complement ESG by providing guidance on what responsible environmental conduct looks like. They encourage companies to proactively manage issues like energy consumption and emissions – which for AI-heavy companies includes the electricity usage of data centers, training models, and device hardware. An example of alignment in action is how tech giants have publicized their AI sustainability efforts: Google, Microsoft, and others regularly report progress on making AI models more efficient and increasing the share of green energy powering their cloud services. These moves are not just for goodwill; they are increasingly demanded by stakeholders. Civil society in Asia, for instance, has begun scrutinizing the rush to build data centres and calling for accountability in ensuring these investments are environmentally sound. “The carbon footprint of technology and AI are legitimate fears… Civil society needs to be asking critical questions about the carbon footprint of data centre investments,” urged one sustainability advocate in Singapore. Such pressure is pushing companies to adopt greener practices or risk reputational and financial repercussions.
Policymakers are also embedding environmental responsibility into digital governance. We see this in Singapore’s requirement of energy assessments for new large-scale computing projects, or in international discussions about adding “sustainability” as a component of AI ethics guidelines. Aligning AI with environmental goals might involve simple steps, like scheduling energy-intensive tasks (e.g., model training) for times when renewable energy is plentiful, or more strategic ones, like R&D investment in energy-efficient algorithms and hardware (as urged by the concept of Green AI). On the governance side, companies might integrate an Environmental Impact Review into AI project lifecycles, mirroring how products are checked for quality and privacy.
The benefit of this alignment is twofold: it reduces the ecological harm of AI deployment, and it also future-proofs the organization. As countries roll out carbon pricing, stricter emissions regulations, or sustainability reporting mandates, companies that have baked environmental thinking into their AI strategy will be ahead of the curve. Moreover, there is a clear ethical dimension – consistent with ISO 26000 – that a truly socially responsible AI is one that respects planetary boundaries and contributes to global sustainability efforts. In summary, incorporating ESG thinking into AI development ensures that innovation does not outpace responsibility. By doing so, organizations uphold the trust of their stakeholders and meet the rising bar for what is considered acceptable corporate conduct in the age of climate change.
AI’s relationship with the environment is a complex balancing act. On one hand, AI contributes to environmental strain through its energy hunger and resource needs; on the other hand, it offers unprecedented tools to monitor ecosystems, fight climate change, and optimize how we use resources. The challenge and opportunity for businesses, governments, and society is to tilt this balance in favor of sustainability. Standards like ISO 26000 remind us that technological advancement should go hand-in-hand with ethical, responsible practices – the environment being a core pillar of responsibility.
In Asia, where the stakes are high due to fast growth and acute climate vulnerabilities, we are seeing the early moves of this balancing act: Singapore’s policies ensuring data centres are efficient and green, corporate initiatives to power AI with clean energy, and innovative AI-driven projects tackling environmental issues. These examples set a positive precedent. They indicate that with conscious effort, AI can be developed in a way that respects environmental limits and even helps to restore and protect our natural world.
As we move forward, it will be crucial for more organizations to integrate “green thinking” into AI design and deployment from the outset – not as an afterthought. This means interdisciplinary collaboration: engineers working with sustainability experts, policymakers providing the right incentives and regulations, and consumers/users being aware of the hidden environmental costs of their digital activities. If successful, the concept of Green AI will become the norm: AI systems that are efficient by design, powered by renewables, and applied to solve sustainability challenges.
Ultimately, AI and environmental sustainability need not be at odds. Guided by CSR principles and robust frameworks like ISO 26000, the tech community in Asia and beyond can ensure that AI becomes a force for good in the climate fight. By doing so, we safeguard the environment even as we unlock AI’s transformative potential – ensuring that progress in AI supports a healthy planet rather than undermines it. This responsible path is not only a moral imperative but also a practical one, as our future economic prosperity and well-being are deeply intertwined with the health of the environment. In the journey of AI and CSR, embracing environmental stewardship will be key to building a resilient and sustainable digital future.
Q: What is ISO 26000 and why is it relevant to AI and the environment?
A: ISO 26000 is an international standard providing guidance on social responsibility (including environmental care). It’s relevant to AI because it outlines principles (like reducing pollution and resource use) that help ensure AI systems are developed and used in an environmentally responsible way.
Q: Why does AI have such a large carbon footprint?
A: Training and running AI models require intensive computation, which consumes a lot of electricity. If this electricity comes from fossil fuels, it results in significant carbon emissions. Additionally, data centers powering AI need energy for both computing and cooling, all of which contributes to AI’s carbon footprint.
Q: What are “green data centres” and how do they help?
A: Green data centres are data storage and processing facilities designed for maximum energy efficiency and minimal environmental impact. They might use advanced cooling technologies, efficient hardware, and renewable energy. By reducing the power and water needed to run servers (which AI relies on), they help lower the overall emissions and resource use of AI services.
Q: How can AI be used to fight climate change or help the environment?
A: AI can analyze vast amounts of data to find patterns and optimize systems in ways humans can’t easily do. For example, AI helps integrate renewable energy into power grids, improves energy efficiency in buildings and transportation (like smarter traffic light control), predicts environmental risks (such as floods or wildfires), and aids in conservation efforts by monitoring ecosystems. These applications of AI support climate action and environmental sustainability.
Q: What steps are companies in Asia taking to make AI more sustainable?
A: Companies and governments in Asia are adopting several strategies: investing in energy-efficient and green data centres, powering operations with renewable energy, optimizing AI algorithms to be less energy-intensive, and following sustainability frameworks (like ISO 26000 or ESG criteria). In Singapore, for instance, new standards and roadmaps guide data centres to reduce their carbon and energy footprint, and firms are piloting AI solutions that both improve business outcomes and contribute to environmental goals.
