The role of Green AI in Singapore’s digital sustainability journey

The benefits of digitalisation are endless for businesses, from increased efficiency to innovation. In fact, businesses have realised this and the digital revolution is already underway. The adoption rate of artificial intelligence (AI) is on the rise, with larger enterprises leading at 44% and SMEs at 4.2%¹ . But as AI adoption continues to grow, so does its environmental footprint. This has sparked a critical question worldwide – can we keep powering innovation without further compromising our planet? Being the cornerstone of Singapore’s digital economy, AI optimises everything from public transport routes to healthcare diagnostics and urban planning. However, with every new AI innovation, energy demands increase, and this often results in the Jevons Paradox – a phenomenon where improvements in energy efficiency leads to greater overall energy consumption instead of less. This paradox highlights the needs for Green AI – a path towards the development of energy-efficient AI systems powered by low- or zero-carbon energy sources². At the end of the day, the dual narrative between AI innovation and environmental stewardship should not be a contradiction; they should mutually reinforce one another. 

Achieving this harmony is not solely the responsibility of one stakeholder but rather, a collaborative effort across the AI value chain, including developers, businesses, and policymakers. Recognising this, IMDA has been actively facilitating and joining global conversations around digital sustainability. In October 2024, IMDA hosted the inaugural Singapore edition of the Green Software Foundation (GSF) Global Summit, a gathering of key players to discuss a sustainable digital future and the role that green software and AI have to play in it. This edition was the only one hosted by a government agency. Such proactive efforts demonstrate Singapore’s commitment to ensuring digital expansion aligns with the nation’s sustainability goals and global leadership. Through collaboration, pilot programmes, and robust partnerships, Singapore is setting an example in the region, with IMDA’s initiatives providing a blueprint for sustainable growth. 

Unpacking Green AI’s impact and its advancements

Green AI, with its potential for energy efficiency, minimal carbon imprint, and responsible resource management, is more than a buzzword³. It is a strategic asset for economies, industries, and enterprises looking to innovate without harming the environment. For these entities, Green AI means rethinking how to handle data, design software, and build infrastructure, while reducing their carbon footprint. 

This strategy is well-timed. With Singapore’s Green Plan 2030 and the goal of net-zero emissions by 2045, adopting sustainable AI practices is no longer optional. This shift is particularly crucial in data centres. While they are essential to powering AI and digital services, data centres are high energy consumers with a hefty carbon footprint. In response to this challenge, Singapore introduced the world’s first sustainability standard designed specifically for data centres in tropical climates. Developed by IMDA and industry partners, this standard sets a benchmark for energy efficiency, renewable energy adoptions, and operational resilience, providing businesses with vital guidance and support to navigate this new terrain. This initiative is further supplemented by the Green Data Centre Roadmap; a blueprint for data centres to reduce their environmental impact while supporting the nation’s growing digital economy. Beyond data centres, Green AI applications are already finding their way into Singapore’s economy, from predictive maintenance in manufacturing to energy-efficient smart grids and data centres. As a forward-thinking, tech-driven city-state, implementing such energy-efficient AI systems, powered by renewable sources, represents a significant step towards lowering the nation’s carbon footprint. 

To truly harness the power of Green AI, businesses need to make it a core focus within their organisation. One way to achieve this is by leveraging ICT for Green, where digital solutions play a critical role in helping businesses from other sectors reduce their carbon footprint. This has the potential to lower global emissions by 20% by 2050 in the three highest-emitting sectors: energy, materials, and mobility⁴. For example, data analytics can optimise energy usage in buildings, transport routes can be refined to cut fuel consumption, and simulations can eliminate waste by replacing physical building prototypes. 

Building on this foundation, companies can adopt three main strategies to pursue their Green AI journey⁵:

1. Greening systems
   This involves the use of energy-efficient infrastructure that minimises carbon emissions from the ground up. By adopting cloud-based solutions to optimise energy use, and transitioning to renewable energy sources for power needs, companies can significantly reduce the environmental price tag on their digital transformation operations.

   That said, it’s important to ensure that cloud-based solutions are well-implemented and optimised, as inefficiencies in deployment can diminish their environmental benefits. Selecting net-zero infrastructure providers and leveraging efficient hardware solutions are additional steps in building truly sustainable systems.

2. Greening data
   Smart data management is vital in building a sustainable AI foundation. In tropical climates like Singapore, cooling systems in data centres, which can constitute up to 40% of the centre’s total energy consumption, pose a major challenge.

   To address this, businesses can adopt innovative solutions like liquid cooling, as well as shift towards a more prudent data approach, by reducing data redundancy, consolidating data storage, and employing energy-friendly data processing methods. Moreover, companies that use AI to process data on-demand rather than inconstant cycles are better able to save energy and operational costs.

3. Greening intelligence
   This strategy looks at designing algorithms that prioritises sustainability without sacrificing performance. By considering processing time, energy use, and data requirements, these algorithms ensure AI is used responsibly and efficiently. AI models operate in two phases: training, where they learn by processing vast amounts of data (20% of their environmental footprint), and inference, where they solve real-world problems (80% of their footprint)⁶. As AI expands, the energy demand for inference is expected to surge. Likening opening peanuts with a sledgehammer, running the most powerful models on routine corporate use cases is not only costly, but builds a larger carbon footprint.

   In this context, smaller, task-specific models offer a solution. These models require less power for both training and inference while maintaining high performance. Companies that design and leverage smaller AI models with more specific use cases to execute complex tasks efficiently can reduce their environmental impact without compromising on the outcomes, gaining a competitive edge in sustainable innovation.

By embedding these strategies into their businesses’ DNA, future-forward companies not only contribute to Singapore’s sustainability goals but also unlock the benefits of Green AI. These include cost savings, enhanced operational efficiency, and a strengthened brand reputation as eco-conscious entities.

IMDA: Driving Green AI in Singapore

For Green AI to truly flourish, the impact of AI on energy consumption and carbon emissions must be quantified and understood. IMDA leads this effort; not just in tracking data but in setting industry standards and creating scalable solutions that businesses can put into practice. 

A key initiative is IMDA’s Green Software Trials. In these trials, IMDA invites industry partners to test out energy-saving methods across various AI applications to champion the advancement and adoption of green digital solutions. Imagine experimenting with AI in a real-world setting and figuring out exactly how to make it greener – that is precisely what these trials do. They assess and quantify the impact of carbon reduction techniques, such as resource redistribution, application modernisation, AI optimisation, and computational offload. By building actionable insights and demonstrating the feasibility of Green AI in real-world applications, the trials then provide a roadmap for other businesses to follow. 

Green AI continues to be a consolidated effort, led by partnerships between industry and government leaders. IMDA partnered with Microsoft to develop the Sustainable Software Development Guidelines, a practical framework to help developers build efficient, low-impact software from the ground up. Additional collaborations on local projects, such as the pilot Data Centre – Call for Application (DC-CFA) exercise, as well as participation in the Global Partnership on AI, the World Economic Forum AI Governance Alliance, and the UN High-Level Advisory Body on AI, underscore IMDA’s commitment to sustainable AI. These partnerships also serve as platforms for ongoing dialogue among government agencies, academic institutions, and industry firms to help formulate guidelines and regulations that support innovation while balancing the risks.

Of special note, Singapore – through IMDA – is also the first country to join the GSF. This milestone was highlighted at the recent GSF Global Summit, where key speakers and industry leaders shared insights on sustainable AI practices. Reinforcing Singapore’s leadership in advancing Green AI, the Tony Blair Institute for Global Change acknowledged Singapore as a pioneer in advancing research and policy that supports sustainable digital economic growth, on the sidelines of Asia Tech x Singapore 2024.

Towards a Green AI-enabled digital future

The rapid growth of AI demands a proactive approach to managing its environmental impact. Green AI provides a pathway for businesses to innovate sustainably – harnessing the potential of AI while safeguarding our planet. As companies continue to rely on AI, adopting Green AI practices becomes essential to fostering responsible, forward-thinking growth. 

IMDA remains at the forefront of the digital sustainability journey, guiding businesses, developers, and policymakers towards a greener digital future. Partnering with IMDA will bring access to expertise, resources, and a vibrant community of eco-conscious digital innovators. Embrace Green AI today – the engine of sustainable growth for your business and the nation.