Artificial Intelligence consumes significant amounts of energy, relying on power-intensive data centres and computing infrastructure. In much of the Asia-Pacific region, this energy still comes from fossil fuels, raising concerns about AI’s growing carbon footprint. Training a single large model can generate over 626,000 pounds of carbon dioxide equivalent, underscoring the environmental cost of rapid innovation.
Despite these concerns, AI can play a constructive role in advancing sustainability. When applied responsibly, it helps optimise energy use, reduce waste, and improve efficiency across sectors. From managing smart grids to streamlining logistics, AI offers tools that support more sustainable operations and decision-making.
The challenge now is to ensure AI development aligns with climate goals. Companies must embed sustainability into their AI strategies, invest in cleaner technologies, and establish strong governance frameworks.
We speak with industry leaders and experts to explore whether AI is hindering or enabling sustainability, and what steps organisations must take to lead responsibly.
Tee Jyh Chong, Senior Vice President, Asia Pacific, Alcatel-Lucent Enterprise
Suvig Sharma, Regional Head, Asia, Confluent
Danny Elmarji, Vice President, Presales, APJC, Dell Technologies
Serene Nah, Managing Director and Head of Asia Pacific, Digital Realty
Kristin Moyer, Distinguished VP and Gartner Fellow, Gartner's CIO Research Group
Howie Lau, Chief Corporate Development and Synergy Officer, NCS
Jan Wuppermann, Senior Vice President, Data & AI Asia Pacific, NTT DATA
Vincent Caldeira, Chief Technology Officer, APAC at Red Hat
Susanna Hasenoehrl, Head of Sustainability Solutions, SAP Asia
Frederic Godemel, Executive Vice President, Energy Management Business, Schneider Electric
Futoshi Niizuma, Vice President for Asia Pacific and Japan (APJ), Seagate Technology
Daniel Pointon, Group Chief Technology Officer, ST Telemedia Global Data Centres
iTnews Asia: How has AI impacted your organisation or business in APAC? How is your organisation/business using or harnessing the potential of AI? Are there specific AI technologies that are being used to reduce energy consumption or environmental impact?
Moyer (Gartner): AI is rapidly reshaping industries and economies worldwide, including the APAC region. CEOs identify AI as the technology that is most transforming their industry.
AI is optimising electricity, water and waste operations, which improves resource efficiency in buildings, industries and cities. Finance departments are using AI to simplify and expedite carbon accounting. Legal teams are using AI to audit for environmental sustainability compliance in contracts. AI is supporting the circular economy by improving product design, tracking and reporting and end-of-life engagement.
AI has the biggest impact on reducing environmental damage when combined with other technologies. Robotic process automation (RPA) and enterprise resource planning (ERP) software with integrated AI are improving manufacturing. Smart grid solutions are enabling data centres to stabilise energy demand and feed stored renewable power back into the grid. AI, IoT and blockchain are tracking the palm oil supply chain.
Lau (NCS): AI has transitioned from a research-driven field into a commercially viable, industry-defining force, reimagining domains including public service, healthcare, finance, transport, and telecommunications.
Many of our solutions have helped our clients work more efficiently and sustainably. In Melbourne, NCS Australia developed a cloud-based digital twin model providing real-time monitoring and analysis of water management parameters. By combining historical data and integrating data from multiple platforms, we can forecast the quality of recycled water using machine learning algorithms.
In Singapore, the NCS AI Centre of Excellence team has worked alongside local industry research partners to leverage AI and Machine Learning, IoT to reduce the carbon footprint of our data centre by improving its cooling performance and energy efficiency.
Nah (Digital Realty): (For data centres), AI has redefined requirements, demanding that infrastructure can handle increasingly dense, high-performance workloads without compromising sustainability. We don’t view this as a disruption, but rather as an opportunity to build smarter and more efficient infrastructure.
Digital Realty prioritises energy efficiency through direct liquid cooling, which efficiently dissipates heat by circulating a liquid coolant directly to heat-generating server components. We also developed Apollo, our proprietary AI platform, which uses operational data to optimise energy usage across our global portfolio. Apollo has already identified approximately 18 gigawatt-hour (GWh) of savings across 12 sites, with 14 GWh implementable immediately.
By architecting for energy efficiency at both workload and infrastructure levels, we ensure that AI performance and sustainability scale hand-in-hand.
Caldeira (Red Hat): AI impacts enterprises, with many struggling to scale AI workloads and transition generative AI from pilot to production, requiring resilient infrastructure and good developer experiences. As a founding member of the global AI Alliance and a member of AI Verify Foundation in Singapore, we aim to democratise AI through open-source contributions, enabling efficient deployment across hybrid cloud environments.
Enterprises are leveraging AI for hyper-automation to streamline operations and optimise resource management, addressing challenges in sectors such as financial, material science, or healthcare
For sustainability, we work with enterprises to leverage specific AI technologies, in areas such as enhancing hardware utilisation and energy efficiency, and providing real-time energy consumption monitoring for AI workloads.
Hasenoehrl (SAP): We believe the full promise of AI can help to increase enterprise productivity by up to 30 per cent, and new agentic AI solutions can automate time-intensive tasks to free up staff to make better decisions and do valuable work. SAP’s data centres are running on 100 percent green electricity as part of 2030 Net Zero commitment.
We also believe AI should be part of the solution. In fact, AI is helping us to increase cloud data centre energy efficiency, and we are helping customers find the most efficient LLM for their needs, which often means the one that uses the least amount of energy.
There are a range of solutions that we now offer that use AI capabilities to help reduce environmental impact, including identifying carbon emissions for purchased goods or materials, transforming fragmented information into structured insights for sustainability models, validating sustainability certificates, and providing safety instructions.
- Susanna Hasenoehrl, Head of Sustainability Solutions, SAP Asia
Sharma (Confluent): Confluent walks the talk when utilising AI smartly, by integrating real-time data streaming with generative AI, LLMs, and advanced analytics, supporting a wide range of enterprise AI use cases.
To accelerate AI innovation with the successful deployment of AI and advanced analytics, businesses must operate on trusted, real-time data streams that ensure the ability to experiment, scale and innovate with greater agility.
The foundation of data streaming drives Confluent’s ability to curate and stream relevant, high-quality data to AI models - what we call ‘data in motion - thereby reducing the volume of data processed and stored, cutting down sheer computational power for AI training and inference, especially in large-scale generative AI deployments.
Pointon (STT GDC): AI is reshaping the data centre landscape and a core pillar of the company’s growth strategy. STT GDC is proactively designing and delivering purpose-built, AI-ready data centres and adapting existing facilities across key markets.
The rise of AI workloads brings both opportunity and complexity. Meeting the substantial power and cooling requirements of advanced AI infrastructure necessitates more than just deploying specialised hardware such as GPUs and accelerators. It is driving a shift toward more advanced cooling systems to better manage the increased thermal load.
AI is also transforming STT GDC’s internal operations. The company has several active programs ranging from GenAI deployment in our corporate environment to AI-enhanced cooling control systems, which utilise reinforcement learning to reduce energy consumption in our data centres. Initial results indicate energy savings of 10 percent, with the potential to reach up to 30 percent savings in our cooling energy consumption.
Chong (ALE): We use AI not only to optimise our internal operations and activities, but also to offer advanced functionalities in service of our customers.
We took a proactive approach to AI integration well before the ChatGPT revolution. The company has an internal charter that guides how we select and deploy AI services across solutions, ensuring we consider not just operational performance and cybersecurity, but also regulatory compliance, ethics, and environmental impact.
- Tee Jyh Chong, Senior Vice President, Asia Pacific, Alcatel-Lucent Enterprise
We often favour a traditional machine learning model over a large language model if it can deliver the same results efficiently, and are also investing in optimised fine-tuning methods to minimise memory usage and computational requirements. Our AI-powered smart building solutions provide the digital foundation needed for sustainable operations.
Through automated systems that integrate lighting, HVAC, and other building services with intelligent sensors, we can ensure energy is only used when and where it is needed. The result is buildings that automatically optimise their environmental footprint while maintaining optimal conditions for occupants.
Elmarji (Dell): We believe the road to efficient AI implementation starts with a focus on efficient infrastructure, and are investing in hardware designed for optimal performance per watt. We look at solutions that leverage advanced cooling technologies, like direct-to-chip or rack-level liquid cooling. The efficiency of hardware, combined with advanced power management tools and thoughtful data centre design, plays a crucial role in reducing energy consumption.
We use AI to automate the collection and analysis of usage data and data centre operations (energy consumption, emissions, waste generation), making sustainability reporting more accurate, efficient, and auditable. This helps us track progress against sustainability goals, optimise the IT environment and identify areas for improvement whilst driving down operational costs.
Godemel (Schneider): AI is accelerating our mission to drive efficiency and sustainability at scale. The true transformative power of AI lies in its ability to drive decarbonisation. From predictive maintenance to intelligent energy optimisation, AI enables our customers to reduce waste, enhance operational efficiency, and cut time, cost, and carbon emissions.
A prime example is our EcoStruxure Microgrid Advisor, which leverages AI to manage distributed energy resources. At our own Boston R&D hub, it orchestrates energy flows from over 1,400 solar panels and energy storage systems, supporting EV charging while ensuring operations remain energy-efficient and resilient.
Internally, we’ve embedded AI across core functions throughout the business – including sales, customer service, marketing, and R&D – to improve productivity, decision-making, and remove repetitive tasks.
AI is foundational to our vision of Electricity 4.0 – where electrification and digitalisation come together to build sustainable, efficient, and resilient business models. We're helping customers apply AI across areas like environmental monitoring, smart grid management, and ESG reporting.
Wuppermann (NTT DATA): We help drive productivity and efficiency in our clients operating model and environment by embedding AI based solutions and governance processes.
We're also embracing the role of Client Zero by leading with our own transformation – truly walking the talk. This includes reimagining our delivery model across application development and embedding AI deeply into our Managed Services platforms, and leveraging built-in AI capabilities from leading hyperscalers and software providers.
To reduce energy consumption, we’ve developed a global AI-ready strategy for all our data centres – whether through building new, greenfield facilities equipped with advanced technologies like liquid cooling, or by upgrading our existing data centres to meet AI readiness standards.
Generative AI has become a catalyst for transforming our services, enabling cognitive-process integration that unifies recognition, memory, and decision-making into seamless workflows.
Futoshi Niizuma (Seagate): AI is transforming how data is generated, stored, and scaled. In turn, businesses are required to rethink their data centre infrastructure. We see this evolution up close: as AI adoption accelerates, the demand for high-performance, energy-efficient storage is growing at an unprecedented rate.
AI also plays a key role in how we operate. We’ve implemented a wide range of AI tools such as generative AI for software coding, autonomous sensor data monitoring, machine learning, automated vision system enhancement, and more smart manufacturing projects globally.
We leveraged AI to enhance precision, improve yields, and reduce energy and material waste. How our AI-driven solutions tackle challenges like anomaly detection, data cleaning, and real-time inferencing is a clear example of how AI can be a catalyst for sustainable innovation without compromising performance.
These insights feed directly into our product design.
For instance, our next-generation storage technologies now deliver up to 3 times more capacity in the same physical footprint while significantly reducing power consumption and embodied carbon per terabyte. AI is not just reshaping our operations; it’s redefining what sustainable infrastructure looks like.
- Futoshi Niizuma, Vice President for Asia Pacific and Japan (APJ), Seagate Technology
iTnews Asia: In APAC or your country/market, what challenges (economic barriers, difficulties across the supply chain, etc.) do businesses face in using AI in ways that do not harm the environment or create waste?
Hasenoehrl (SAP): The biggest challenge – and opportunity – for many businesses across Asia Pacific adopting AI is data. Good data makes good AI. Without it, we risk wasted energy, time, money, and emissions. That’s not good for business or the environment.
Real, sustainable outcomes require the use of real sustainability data, not averages or estimates. And once that data is identified and automated, sustainability data should be integrated at the very core of business, in every business decision concerning product design, business process, and functional practice. Only then can businesses make decisions in a responsible, transparent, and auditable way – and use that data to fuel AI applications that improve efficiency and productivity.
Wuppermann (NTT DATA): Rapid innovation in AI often outpaces the development of frameworks for ethics, safety, sustainability, and inclusiveness, creating a ‘responsibility gap’ that exposes businesses to risks such as wasted investment and delays in moving from experimentation to full-scale implementation. C-suite leaders in APAC face tension between prioritising breakthrough innovation and ensuring accountability.
Additionally, the absence of unified standards for transparency and sustainability forces each organisation to define its metrics, which can slow collaboration and inhibit industry-wide progress. Increased supply chain volatility continues to overshadow environmental concerns in non-regulated industries.
Bridging this gap requires adopting principles such as sustainable development, human autonomy, security, privacy, communication, and co-creation, embedding them from the earliest stages of AI strategy.
Moyer (Gartner):
AI hasn’t yet hit a break-even point where it is helping sustainability more than hurting it. AI-driven applications require immense computing power, and this demands substantial data centre capacity. The addition of AI-driven computing loads exacerbates the strain on energy grids, which in many cases in APAC still rely on fossil fuels.
- Kristin Moyer, Distinguished VP and Gartner Fellow, Gartner's CIO Research Group
In Australia and New Zealand, technology leaders say they don’t know the impact of AI on energy consumption. Very few believe their organisation pays enough attention to the energy consumption of data centres. In Australia, there's a political debate over investing in nuclear vs. renewables to meet power demands.
Niizuma (Seagate): Awareness is rising across Asia, the drive to decarbonise is strong, but practical constraints continue to limit progress. In South Korea and Singapore, the majority of data centre professionals, respectively, cite the limited availability of sustainable electricity as a key barrier. Similarly, in India and Japan, infrastructure limitations remain a common concern.
Financial constraints are another shared challenge. Singapore and South Korea estimate that at least US 5.5 billion is required to transition to more sustainable data storage infrastructure. Taiwan, Japan, Australia and India reported similarly high figures between US 4.5 to 5.1 billion, underscoring the scale of investment needed to modernise legacy systems, adopt renewable energy, and implement circular practices. Then there’s talent, professionals in Singapore, Taiwan and India identified workforce training costs as a major constraint.
These challenges demonstrate that responsible AI adoption isn’t just about deploying new technologies, but building the right systems, skills, and support structures. Progress will require coordinated public-private action to overcome these systemic barriers.
Nah (Digital Realty): Operating data centres in Singapore has its unique challenges owing to land, water and energy constraints. To address this, Digital Realty collaborated with the Infocomm Media Development Authority (IMDA) to pilot a new standard for operating data centres in Singapore, where operating temperatures were raised by 2°C at two of our data halls in Singapore.
The result: approximately 2-3 percent reduction in total energy usage in the data halls over the trial period. We’ve also launched a cooling tower project with Singapore’s Public Utilities Board (PUB), which tripled concentration cycles and resulted in a 650,000-litre monthly reduction in blowdown water.
Beyond accounting for the technical demands of AI, ensuring sustainable AI requires strategic investment, policy support and sustainability standards to be aligned. Singapore’s green AI vision is promising, but actualising this will depend on practical action and a shared long-term vision to balance innovation and environmental responsibility.
Caldeira (Red Hat): In the APAC market, businesses face significant challenges in adopting AI sustainably. The true sustainability challenge for AI lies not in model training, but in inference, the phase where models deliver predictions and value at scale. Inference is a continuous and highly variable process, consuming energy constantly as it scales with utilisation. This often offsets efficiency gains despite technical innovations.
A primary economic barrier is this high energy consumption, particularly during the continuous inference phase. Organisations struggle with limited expertise in optimising AI workloads and a lack of clear sustainability roadmaps. Supply chain difficulties, such as securing GPUs and sufficient energy, are also prevalent, as many data centres are not equipped for AI's immense power demands.
Furthermore, the absence of shared standards and coordinated efforts across government, industry, academia, and open-source communities makes it difficult to implement AI solutions that minimise environmental harm and waste. Many businesses are also unaware of green software benefits or lack internal expertise.
Lau (NCS):
Economic concerns can be real barriers for businesses because sustainable solutions can come at a cost in the short term. However, in the long term, putting in place practices that do not harm the environment is crucial.
- Howie Lau, Chief Corporate Development and Synergy Officer, NCS
Additionally, businesses need to transition to renewable energy to operate and power computing resources for AI workloads. The Singapore Government is proactively pursuing an ASEAN energy grid. In terms of supply chains, businesses can acquire IT products with low embodied carbon to power their AI systems on-premise.
NCS has launched the Green Code Foundry, a research project to pilot carbon intensity monitoring tools in our internal cloud environment. The tools from the pilot project will help NCS developers identify inefficient codes during the development phase, as well as areas for optimisation of software carbon efficiency.
Pointon (STT GDC): Electricity demand has increased in Southeast Asia, with AI being one of the factors putting added pressure on existing grid infrastructure.
Integrating renewable energy while increasing electricity output is a complex problem that energy providers, grid operators, and the data centre sector need to solve. For example, extending high-voltage power transmission and substation capacities to enable new development is costly and takes many years or more, notwithstanding that the further you transmit electricity, the less efficient it is.
Supply chain constraints add to these challenges. High global demand for specialised electrical components, transformers and advanced cooling systems can lead to procurement challenges.
Regulatory fragmentation presents another challenge. The lack of standardised environmental frameworks or consistent incentives for sustainable AI deployment. The ASEAN Power Grid initiative exemplifies this issue: cross-border renewable energy transmission capabilities are still lagging behind.
Businesses are facing a critical shortage of qualified personnel, from construction workers and engineers to technicians and business professionals. This shortage is also compounded by an aging workforce.
Sharma (Confluent): In spite of Singapore’s government driving a sustainability-focused regulatory framework, there remain barriers to using AI sustainably, due to infrastructure, cost, and skills concerns.
The challenge starts with AI workloads, especially training large models, which require vast computing power, leading to high electricity and water use, particularly acute in Singapore’s tropical climate, where cooling data centres is energy-intensive. Data centres already account for about 7 percent of Singapore’s total electricity consumption, with AI accelerating demand further.
Organisations with legacy systems, especially SMEs, struggle with the capital required to modernise IT infrastructure completely, let alone deploy the best energy-efficient hardware. With cost pressures rising worldwide, organisations typically look to prioritise profitability in the short term, rather than sustainability in the long run.
There is a shortage of talent with the expertise to develop, deploy, and maintain energy-efficient AI systems. Finally, with limited awareness of newer, energy-efficient technologies and green AI principles among business leaders and IT staff, coupled with a lack of defined ROIs around ‘green’ AI, these deter businesses from prioritising sustainability over short-term gains, further impeding progress.
Elmarji (Dell): A key consideration for companies as they consider AI is the power demands associated with big training workloads and large language models.
Some experts predict data centre energy consumption could double by 2030, placing added strain on already burdened power grids. Is the present energy infrastructure equipped to meet that demand? Reliable, resilient and affordable energy has become a top priority for data centre operations.
- Danny Elmarji, Vice President, Presales, APJC, Dell Technologies
It is also important not to confuse the power requirements of large training versus the inference environment of the enterprise. In most early enterprise adoption cases, the sustainability impact and energy demands of the tech deployed are not as outsized as you might assume, and would thrive well within the existing power budget in data centres and PCs.
Improving energy usage requires collaboration across the tech ecosystem. To support the sustainable growth of AI, we must modernise energy infrastructure, invest in diverse energy sources, and incentivise the development of energy-efficient data centres.
Godemel (Schneider): APAC is a diverse market, home to both advanced digital economies and fast-growing infrastructure-constrained markets. Businesses of all sizes face a range of challenges in deploying AI sustainably, from high upfront investment costs, limited technical expertise, regulatory uncertainty, and fragmented supply chains.
One of the biggest challenges in the region is managing the energy footprint of AI itself. As adoption accelerates across sectors – from manufacturing to smart cities – the demand for computing power, data storage, and connectivity rises sharply, putting added pressure on energy systems. In markets still transitioning away from fossil fuels, this can lead to unintended increases in emissions and energy waste.
Data maturity is another hurdle. Many businesses still face fragmented systems and limited access to quality data, essential for AI to deliver actionable insights.
Economic and structural constraints also complicate adoption. In emerging markets, the cost of sustainable AI infrastructure can be prohibitive. Public-private partnerships, financing support, and government incentives are often needed to make these solutions viable.
Chong (ALE): Many APAC markets still depend heavily on fossil fuel-powered grids, which means even the most efficiently designed AI systems are running on dirty energy. As such, this creates a major obstacle for businesses looking to adopt AI in a responsible manner.
Economic pressures also play a greater role in overriding sustainability intentions. Companies, particularly SMEs, face genuine dilemmas when energy-efficient AI solutions require higher upfront investments, even when they deliver long-term savings; the business case for sustainability becomes harder to justify when cash flow is tight.
The region's vast geography compounds these challenges significantly. Procuring energy-efficient hardware often involves complex supply chains spanning multiple countries with varying infrastructure capabilities. These extended procurement cycles not only delay deployment but ironically increase the carbon footprint of the very solutions designed to reduce environmental impact.
Most critically, there is a substantial knowledge gap. Many organisations lack the technical expertise to optimise AI deployments effectively, leading to inefficient implementations that consume more resources than necessary.
iTnews Asia: Is there enough awareness in APAC on the importance of sustainable AI? What role should governments play in ensuring AI development aligns with environmental goals (need for standards, incentives, or regulation beyond corporate initiatives)?
Caldeira (Red Hat): In APAC, businesses lack sufficient awareness of green software benefits and internal expertise to implement sustainable AI practices. Despite AI's increasing societal integration, the conversation around its sustainability is still evolving.
Governments play a crucial role beyond corporate initiatives. They should set clear standards, offer incentives, and create "safe areas" or sandbox environments for testing green AI solutions. Singapore serves as a strong example, fostering partnerships between government, industry, and academia to drive innovation.
- Vincent Caldeira, Chief Technology Officer, APAC at Red Hat
Emerging regulations, like the EU AI Act, mandate disclosure of AI system energy consumption and environmental impact, a trend expected to become global. Governments should also promote open-source software, which reduces energy consumption through reuse and transparency, preventing redundant efforts and ensuring trusted AI models. Ultimately, a partnership-driven model involving governments, businesses, and academia is essential.
Wuppermann (NTT DATA): Awareness of sustainable AI is growing, albeit from a small base across APAC, but remains uneven. APAC IT leaders consider current government regulations on GenAI to be unclear, which stifles innovation and discourages investment; most expect GenAI-related compliance spending to rise as a result.
Governments should establish clear, region-wide standards for energy efficiency, carbon reduction, and resource management so companies can benchmark, measure, and disclose their performance against shared metrics.
Providing grants or subsidies for AI research focused on energy optimisation and renewable-energy integration will incentivise the adoption of greener AI technologies.
By clarifying regulations, providing incentives, and enforcing minimum-efficiency standards, policymakers can bridge the current responsibility gap and drive collective progress toward sustainable AI.
Governments should play a more active role in defining the level playing field and minimum required standards. This is still not driven fast enough particularly when compared to the speed of AI technology evolution and sustainability requirements to adhere to the Paris Agreement.
Sharma (Confluent): Awareness of AI’s environmental impact has risen. However, the understanding and implementation of sustainable AI practices remain fragmented. This gap stems from many organisations still focusing on how to fully harness the potential of AI while dealing with cost pressures – let alone its broader societal and environmental implications.
Governments can play a more active role by fostering public-private partnerships. These partnerships can co-create the physical and digital infrastructure that supports both innovation and sustainability in AI, while establishing region-specific sustainability benchmarks for AI workloads.
We need broader ecosystem collaboration to drive sustainable AI at scale. Governments can play a more active role in driving greener AI development, shaping how AI development aligns with environmental goals - particularly through fostering public-private partnerships.
These partnerships can co-create the physical and digital infrastructure that supports both innovation and sustainability in AI — while establishing region-specific sustainability benchmarks for AI workloads.
Pointon (STT GDC): Awareness of sustainable AI in APAC is on the rise, but it remains uneven across the region. There are encouraging signs of progress. For example, Malaysia has committed over US$1.5 billion to digital infrastructure, which has resulted in a tenfold market expansion.
Thailand’s establishment of special economic zones, such as the Eastern Economic Corridor, allows for 100 percent foreign ownership of data centre investments within designated areas.
Singapore’s approach the impact of public-private partnerships with initiatives like the National AI Strategy 2.0 and efforts to develop carbon import pathways with neighbouring countries providing frameworks for sustainable AI development.
Governments need to accelerate infrastructure investments and regulatory harmonisation. Effective policy requires a careful balance between rapid AI adoption and upholding sustainability mandates.
Reducing red tape around sustainable infrastructure projects, implementing standardised environmental reporting frameworks, and providing incentives that make green AI deployment economically attractive as well as environmentally responsible would significantly advance the cause.
- Daniel Pointon, Group Chief Technology Officer, ST Telemedia Global Data Centres
Godemel (Schneider): Awareness of AI’s environmental impact is rising across APAC, but adoption remains uneven. While some organisations are actively deploying AI to support their decarbonisation goals, others are still navigating the early stages – building understanding, capabilities and infrastructure.
As AI usage scales, so too does the need to manage its resource intensity – from computing power, data storage, and energy consumption.
AI adoption must be matched by commitment to responsible, low-impact growth. Achieving this balance requires more than corporate action. Governments across APAC play a critical role in setting direction – by establishing clear standards, offering regulatory certainty, and enabling the right incentives to ensure AI is powered by decarbonised energy source as much as possible.
Tools like green AI certifications, carbon disclosure mandates, and public investments in sustainable infrastructure will be essential in aligning innovation and with environmental responsibility.
Niizuma (Seagate): Awareness is rising across Ais, but meaningful action still trails behind. From our research, every respondent surveyed in South Korea and Japan expressed concern about the environmental footprint of data infrastructure. Similarly, a clear majority across the wider APJ region also shared this sentiment.
Yet these concerns rarely shape procurement decisions. In Singapore and India, fewer than 17 percent said environmental impact influences purchasing. This disconnect reflects a broader regional gap between intent and execution.
While we see governments beginning to respond, such as Singapore’s National AI Strategy 2.0, Japan’s Green Growth Strategy and growing interest in the International Sustainability Standards Board (ISSB) framework across APJ, reporting alone won’t be enough.
Public policy needs to support infrastructure upgrades, incentivise low-emission AI models, and foster a robust ecosystem for carbon measurement and skills development. From our perspective, real progress depends on better alignment between national ambitions, regulatory standards, and the practical capabilities companies need to comply and thrive.
Chong (ALE): Awareness of sustainable AI in APAC is growing but remains uneven across the region; SMEs often view sustainability as secondary to growth. That said, there is still room for effective policy to balance rapid AI adoption with sustainability mandates. The region would benefit from comprehensive AI regulation like the EU's AI Act, which primarily focuses on protecting end-users from AI risks but could be expanded to include mandatory sustainability obligations.
This would be a logical extension given that AI's environmental impact indirectly affects public health and human welfare. Investment in regional renewable energy infrastructure is crucial, especially in developing APAC markets.
Just as important are government mandates for sustainability criteria in public sector AI procurement and mandatory carbon reporting standards for AI systems.
A comprehensive regional AI regulation combining both safety and sustainability mandates would provide the necessary policy framework to ensure AI development serves technological progress while protecting human welfare.
The private sector also needs to play its part in leading by example.
Moyer (Gartner): It is very difficult for organisations in APAC to know what their AI-related emissions are. This makes it impossible to have full awareness of the impact of AI, both good and bad.
Governments can support AI and sustainability by taking a proactive approach to clean energy for data centre infrastructure through subsidies and tax incentives.
Governments can set sustainability targets for digital infrastructure. For example, Japan's Green Growth Strategy requires data centres to use renewable energy for a portion of their energy requirements. Singapore lifted its data centre moratorium and then put more stringent energy, water and environmental impact standards in place.
Governments can collaborate across borders. Asia Zero Emission Centre is a collaboration centre across countries that is trying to serve as a platform for policy dialogues, knowledge and innovation.
Lau (NCS): Most countries in APAC have announced measures to reduce carbon emissions through carbon pricing instruments. At the national level, the Singapore government has introduced a carbon tax to encourage all sectors of the economy to reduce greenhouse gas emissions and invest in sustainable practices. This sends a strong signal to businesses and organisations. At the same time, there is a need for collaboration across private and public sectors to raise awareness and share best practices for greening the industry.
For example, government agencies like GovTech Singapore and Infocomm Media Development Authority (IMDA) are among the first to join the Green Software Foundation (GSF), which focuses on building more sustainable AI solutions.
Nah (Digital Realty): Awareness of sustainable AI is growing in APAC, and customers are keen to innovate with AI but require scalable solutions. Governments continue to play a key role in setting the pace and the direction of sustainable AI. Singapore’s Green Data Centre Roadmap is a strong example of how the government is charting a clear direction for the industry to follow.
There are still exciting opportunities to accelerate the green AI transition, such as access to power purchase agreements to implement energy-efficient systems. To accelerate progress, public-private partnerships are key, along with expanded access to renewable energy and efficient technologies.
- Serene Nah, Managing Director and Head of Asia Pacific, Digital Realty
Hasenoehrl (SAP): Awareness of the need for sustainable AI is increasing across APAC.
AI will only be useful if it can be trusted to deliver high levels of security, privacy, compliance, and ethics, as well as managing and mitigating the negative environmental impact. To enable this connection, we need to ensure that AI itself is trustworthy and safe, that the data centres running it are sustainable, and that companies have the consistency to work across jurisdictions.
Continued regulatory clarity and consistency will enable businesses to fully benefit from AI innovations and create meaningful impact for their businesses, economies and the communities they serve.
iTnews Asia: How urgent is it for companies to be responsible in their AI development? How can companies ensure transparency in the environmental impact of their AI systems? (e.g., carbon reporting or lifecycle analysis - reporting mechanisms)
Caldeira (Red Hat): Almost all companies invest in AI, yet only 1 percent believe they have reached maturity. This highlights the urgency for companies to be responsible in their AI practices. AI's increasing energy demands are driving an ‘AI sustainability crisis,’ with global data centre power projected to double by 2026, largely due to AI workloads so incorporating sustainability into AI development is no longer just ethical - it’s essential.
To ensure transparency in AI systems' environmental impact, companies must conduct thorough lifecycle analyses, measuring carbon emissions from hardware production, data centre use, and equipment disposal.
Wuppermann (NTT DATA): Given the accelerating climate crisis, companies must adopt responsible AI practices immediately to maintain stakeholder trust, avoid increasingly stringent regulations, and prevent higher long-term costs.
To ensure transparency in the environmental impact of AI systems, firms should embed sustainability metrics throughout the AI lifecycle. Organisations will need scalable, data-driven “sustainable AI frameworks” that align AI operations with decarbonisation targets.
Report detailed the CO₂ emissions of AI assets. This would also mean that IT buyers should work exclusively with vendors who can demonstrate clear carbon-reporting and lifecycle-analysis compliance, making transparent emissions data a prerequisite for market access.
Integrate sustainability leadership into AI governance. Organisations will embed their chief ethics or sustainability officer within central AI decision-making teams, ensuring environmental reporting is enforced at the executive level. Integrating these metrics into annual sustainability reports demonstrates transparency and drives continuous improvement.
- Jan Wuppermann, Senior Vice President, Data & AI Asia Pacific, NTT DATA
Lau (NCS): Companies must keep up with evolving AI regulations, particularly regarding data privacy, ethical use, and ensure compliance with cybersecurity regulations. This ensures digital resilience efforts align with global standards and reduce exposure to legal and financial penalties.
With regards to carbon reporting and life cycle analysis, how AI systems are reflected in the life cycle assessment will differ among use cases and assumptions, resulting in different calculations for environmental factors and impact.
Pointon (STT GDC): Sustainability cannot be treated as a future consideration. The infrastructure decisions made today will have environmental repercussions for decades to come.
Goldman Sachs Research projects global power demand from data centres will rise to 165 percent by 2030 primarily due to AI workloads.
Ensuring transparency in the environmental impact of AI systems requires comprehensive and standardised reporting mechanisms. In addition to the data centre itself, the specific GPUs or accelerators deployed, as well as the efficiency of their utilisation, have an impact on overall sustainability. Carbon emissions per token will be the new measurement of end-to-end AI efficiency, considering both compute and data centre efficiency.
Companies must integrate carbon accounting into AI project evaluation from the outset. This means tracking not just operational energy consumption but also the full lifecycle impact of AI infrastructure, from manufacturing through to decommissioning.
The most responsible approach combines immediate action—such as partnering with certified sustainable facilities—with long-term commitments to renewable energy sourcing and ongoing efficiency optimisation.
Nah (Digital Realty): AI’s growing footprint demands immediate action. According to forecasts, AI-related workloads could account for about 3 percent of global electricity use by 2030. Responsible infrastructure choices are critical to keep this in check.
Transparency is equally important to help businesses commit to and effect real change. This includes measuring and disclosing the full environmental impact of their AI systems.
We’ve committed to the Science-Based Targets initiative (SBTi) framework, to reduce Scope 1 and 2 emissions by 68 percent, as well as to reduce Scope 3 emissions by 24 percent per square foot by 2030. These measurable targets hold us accountable to real change and guide our day-to-day decisions.
Transparent reporting of digital infrastructure will be the way forward to ensuring sustainable AI growth.
Hasenoehrl (SAP): Responsible AI development is non-negotiable and should be a fundamental part of every AI implementation.
Valuable business AI will be based on the use of mission critical data that comes from the apps enterprises use for key functions like finance, procurement, supply chain, production, and HR, requiring the same stringent safeguards when it comes to security and sovereignty.
Similarly, organisations need to create a comprehensive framework for mitigating bias in AI, ensuring use cases do not cross red lines, such as causing discrimination or abuse.
It is also true we need to ensure full transparency when it comes to emissions across all enterprise activities, including the usage of AI. That means working with AI suppliers and providers to get actual data and integrating into enterprise carbon reporting.
Moyer (Gartner): The rapid growth of data centres, fueled by GenAI technologies, is already exceeding the ability of utilities to provide sufficient power in some locations. Companies cannot fully ensure transparency regarding their environmental impact of AI because most struggle to get adequate scope 3 emissions data from their partners. But they can strive to provide the best data possible.
It’s important to review product carbon footprint (PCF) reports from partners to compare embodied carbon and materials impact. Leverage standard metrics like Power Usage Effectiveness (PUE) to assess data centre energy efficiency, with values closer to 1.0 indicating better efficiency.
Collaborate with AI infrastructure providers and managed service providers to clearly articulate sustainability requirements for all life cycle stages. Develop a well-documented and publicly accessible position on the ethical and responsible use of AI.
Godemel (Schneider): AI can accelerate decarbonisation, optimise energy use, and drive measurable environmental impact. With the right approach, its benefits can far outweigh its footprint. Maximising this potential requires the same discipline applied to any major investment: lifecycle analysis, carbon accounting, and real-time monitoring of energy consumption in AI workloads. Critically, it also means selecting the right model for the right purpose.
Many of today’s most impactful sustainability solutions are powered by analytical AI models that are far less compute-intensive than generative AI. These analytical systems not only deliver precision and speed but often generate energy and cost savings that exceed their operational footprint, often by a factor of hundreds.
Embedding transparency into governance is key. This includes tools like ESG dashboards and AI-specific KPIs, which provide real-time visibility into energy use, emissions, and system efficiency. Boards and executive teams must include environmental impact as a core consideration when evaluating AI applications, ensuring responsible deployment is guided by measurable outcomes.
Niizuma (Seagate): The urgency is real. As AI’s energy and resource demands grow, so does the risk of unintended environmental impact. Transparency is the foundation. Without conscious design, we risk trading intelligence for impact.
Transparency is the foundation. Companies must account for their full environmental footprint, including Scope 3 emissions embedded in infrastructure and supply chains. That means holding suppliers to the same standards we set for ourselves.
At Seagate, our smart manufacturing projects have shown that real-time monitoring of energy use and production efficiency provides the foundation for sustained improvement, not just one-off reporting. This level of operational insight turns sustainability from an obligation into a performance driver.
Sharma (Confluent): It is crucial to start on responsible and sustainable AI development as early in the lifecycle. It incurs a heavier cost on businesses and our world if AI projects are fundamentally energy-intensive and end up being reworked to comprehensively address environmental, ethical, and societal concerns.
Data streaming is a stepping stone towards sustainable AI, providing organisations with comprehensive visibility into their operations, enabling them to respond to events as they happen, not hours or days later. This is especially powerful for green initiatives, where timely action can translate directly into measurable environmental benefits.
Businesses can harness real-time data monitoring to enhance the accuracy and transparency of sustainability reporting. With live insights into energy consumption, emissions, and resource usage, companies can generate precise, up-to-date reports that meet regulatory requirements, create greater value, and build customer trust.
Chong (ALE): The urgency is real, given APAC's rapid AI adoption and its impact on global emissions. We are already implementing Life Cycle Assessment (LCA) practices for our solutions as part of our broader transformation toward circular economy principles.
Achieving transparency in environmental impact requires integrating the digital services component into our LCA frameworks, which means we need comprehensive impact databases that capture the true footprint of AI operations.
However, building these critical databases can only succeed through collaboration across all stakeholders — AI service providers, system integrators, and end users must work together to create the data foundation necessary for accurate environmental assessment of digital services.
Transparency requires robust measurement frameworks; we also track hazardous substances and eliminate banned materials from our products while ensuring compliance with environmental legislation
The key to meaningful progress lies in establishing robust accountability mechanisms that drive continuous improvement rather than simply checking boxes to meet compliance requirements. Companies must honestly assess their environmental impact and commit to measurable improvement over time.
iTnews Asia: What advice can you give to companies on how they can align their AI initiatives with their business and sustainability strategies?
Pointon (STT GDC): Aligning AI initiatives with both business and sustainability strategies is essential for long-term success. The most effective starting point is to forge strategic infrastructure partnerships that deliver immediate AI capabilities while providing a clear pathway to sustainability.
Securing renewable energy partnerships should be an early consideration in the planning process. Experience across the sector shows that sourcing vast quantities of high-quality renewable energy, particularly in developing markets, often requires a planning horizon of three to five years.
Organisations that address energy sourcing at the outset are better positioned to manage costs and ensure their AI operations are powered sustainably. Finally, it is essential to integrate sustainability metrics into the evaluation of AI projects from the beginning, with carbon emissions per token emerging as the definitive end-to-end measurement that captures both computational efficiency and data centre sustainability.
Godemel (Schneider): Start by identifying AI use cases that offer both business value and sustainability benefit. Look for areas where efficiency improvements can also reduce emissions. For example, predictive maintenance can reduce downtime and emissions, while AI-enhanced energy load management in buildings or data centres cuts waste and lowers costs. These aren’t just ideas; they’re already being implemented successfully.
Most importantly, don’t treat AI as a separate tech initiative. Integrate it into your sustainability roadmap from the outset, ensuring that AI solutions are aligned with broader environmental and business priorities. This requires close collaboration between technology, operations, and sustainability teams.
- Frederic Godemel, Executive Vice President, Energy Management Business, Schneider Electric
We’ve seen firsthand how AI can transform operations and sustainability outcomes in tandem. Done right, AI doesn’t just support sustainability, it accelerates it.
Caldeira (Red Hat): To align AI initiatives with business and sustainability strategies, companies must first design their IT infrastructure with sustainability as a foundational principle. A critical step is the ability to consistently measure the energy footprint of software systems in real-time.
By continuously tracking energy consumption and carbon impact, organisations gain the necessary insights to identify inefficiencies and make data-driven decisions for iterative improvements. This transparency ensures a culture of energy-aware computing and allows for active optimisation of AI workloads, leading to reduced environmental impact and cost savings.
Upskilling teams, modernising infrastructure, and fostering cross-functional collaboration are also key to scaling AI responsibly and translating experimentation into lasting, sustainable impact
Moyer (Gartner): Set a clear AI ambition for environmental sustainability. Include the role AI will play, and how the organisation will both mitigate its negative impacts and scale its positive impacts. Clarify sustainability goals and articulate the organisation's sustainability requirements for AI initiatives.
Implement GreenOps to reduce carbon footprint, optimise energy consumption, manage waste and ensure responsible resource use. Prioritise clean and efficient energy for data centres when possible.
Consider running large-scale GenAI processes during off-hours when cleaner energy might be more available and less expensive. Establish green IT policies that promote energy conservation, waste reduction, and responsible e-waste disposal.
Deploy energy-efficient servers, storage, networking equipment, and AI-powered optimisation tools. Educate and train employees by providing training on environmental sustainability and green IT practices.
Nah (Digital Realty): AI is only as sustainable as the infrastructure that it runs on. Companies looking to scale should look at how their data is processed, stored and managed because performance and sustainability are interconnected. This means opting for infrastructure that prioritises sustainability, and that is efficient by design. For example, Digital Realty’s data centers in Singapore run on 100 percent renewable energy
Rather than designing for peak loads alone, businesses should invest in infrastructure that is built for long term sustainability - ensuring both environmental resilience and performance.
Wuppermann (NTT DATA): Companies can embed sustainability into AI governance by defining principles (energy efficiency, ethics, transparency) and having a dedicated oversight team.
They should implement green software practices to measure and minimise software carbon intensity throughout development and deployment. There is also a need to align procurement with ESG criteria by requiring vendors to provide lifecycle analyses detailing carbon emissions from IT assets.
Additionally, invest in workforce training on eco-friendly coding and responsible AI to foster a culture of “responsible reinvention.”
The two biggest challenges we are facing in my opinion are Climate Change and the lack of effective AI Governance. These should be the two non-negotiable elements of each Board agenda (in addition to Cybersecurity) and embedded into every company’s corporate strategy and investment and business plan.
Chong (ALE): Sustainability must be integrated into AI strategy from the get-go. This can start with comprehensive energy audits to establish baseline measurements before AI deployments. Companies can prioritise AI applications according to the value delivered that both relate to business, and environmental improvements.
Companies should also select the right AI service for their specific needs and not hesitate to challenge their providers about the energy consumption of models across different inference scenarios. This means actively questioning providers about computational efficiency, demanding transparency in energy usage metrics, and choosing solutions that balance performance requirements with environmental responsibility.
When evaluating technology investments, companies must expand their analysis beyond traditional financial metrics to include the total cost of ownership, factoring in environmental considerations when evaluating investments in technology and digital infrastructure.
Governance frameworks should be structured in a holistic way to embed environmental impact assessments alongside performance, ROI, security, ethics, sustainability, and reliability. This ensures sustainability considerations influence decision-making at every stage.
Sharma (Confluent): There is undoubtedly a pressing need to become more profitable in our uncertain economy. AI’s energy demands present real sustainability challenges, but businesses don’t have to compromise on innovation to focus on being greener.
Establishing a real-time, trustworthy data foundation allows businesses to capture, integrate and contextualise data from across the organisation, enabling leaders and the workforce to focus on driving both profit and environmental bottom lines.
Innovations that help drive omnichannel personalisation can be used to develop specific workflows that enhance sustainability at every department and level of the organisation - use cases can include predictive maintenance to reduce waste, renewable energy integration and more.
Sustainability considerations should be integrated as much as possible from the outset of AI development. This means evaluating the energy and resource requirements of AI systems, selecting efficient architectures, and designing workflows that minimise waste.
- Suvig Sharma, Regional Head, Asia, Confluent
Niizuma (Seagate): Begin by recognising that sustainability and operational efficiency are increasingly aligned. The same choices that reduce environmental impact – longer-lasting equipment, smaller data footprints, energy-efficient design – also reduce cost and complexity over time.
Second, choose technologies built for scale and efficiency. For instance, higher areal density in storage enables businesses to expand their AI capabilities within the same physical footprint, using less power and fewer materials. These upstream decisions deliver downstream benefits across TCO, carbon reporting, and resource use.
Finally, don’t do it alone. This is an ecosystem challenge. Engage suppliers, customers, and policymakers to build shared standards, circularity programmes, and transparent reporting systems.
In the AI era, performance will increasingly be measured not just by speed or scale, but by sustainability. Getting this right is no longer a competitive edge, it’s a baseline for long-term relevance.
Lau (NCS): Organisations must ensure that technological advancements are matched by strong digital resilience. This is not just about cybersecurity but also extends to safeguarding application performance, infrastructure scalability, data governance and operational responsiveness. The intersection of AI and digital resilience presents both opportunities and risks, requiring a proactive approach from leadership.
A resilient organisation with a sustainability agenda must be adaptive, and education is critical. Leaders should drive AI literacy across all levels, helping employees understand how AI supports resilience, enabling faster recovery from disruptions. Emphasising and including sustainability in these discussions will encourage a culture that sees it as part of the long-term resilience and strength of the business.
Hasenoehrl (SAP): AI offers enormous opportunities to drive not only productivity and revenue growth, but sustainability action. Today, over half of sustainability practitioners plan to improve data analysis using AI.
Using high-value data to drive AI, businesses can reduce cost, lower risk, and drive performance so they can enable faster, smarter decisions for ESG reporting, carbon initiatives, circularity, and compliance. AI will relieve sustainability teams of time-consuming data analysis as well as providing regulatory and supply chain guidance so teams can focus on strategy, implementation, and impact.
At SAP, we can see AI helping customers to generate sustainability reports for internal stakeholders, reducing time-consuming and error-prone manual reporting efforts and processes so teams can focus on adding value. It can help to analyse complex, difficult-to-understand regulations and providing compliance guidance and direction to stay ahead of new requirements. It can also help to automate the calculation of product carbon footprints.
Elmarji (Dell): Building the right infrastructure is a start, as significant savings and performance gains can be achieved by working with experts to tailor your AI solutions and optimise workloads. Businesses can enhance efficiency by adopting a "right-sized" approach to AI.
While some organisations may benefit from large, general-purpose language models, many only need tailored solutions specific to their domain or enterprise. Leveraging pre-trained models, rather than building from scratch, simplifies adoption and significantly reduces energy demands.
Partnering with experts who can customise AI solutions to suit specific workloads is essential for minimising energy waste. Equally important is working with providers that offer asset recovery services to ensure the responsible disposal of outdated hardware.
