AI to play a bigger role in solving Asia’s sustainability challenges: Google

The tech major has been investing heavily in the region to leverage digital growth opportunities, and encouraging local sustainability ventures to use its advanced software tools.

Google singapore data centre
Google has four data centres in Singapore, with the latest sporting sustainability features such as lower water consumption and more efficient cooling. Image: Google.

Asia has faced a bruising few years of climate disasters, with the El Niño weather phenomenon bringing searing heatwaves and droughts interspersed with stronger and more unpredictable typhoon seasons.

The region, which sweltered through the hottest year on record in 2023, is poised to continue bearing the brunt of climate change – exacerbated by rising global temperatures. 

Technology leader Google believes it can help the region’s climate issues through its latest artificial intelligence (AI) tools. The company has made billion-dollar investments to scale up its data and machine learning infrastructure in Singapore, with similar plans slated for Malaysia. It also provides mapping, data processing, and weather prediction software for local groups to better protect nature and livelihoods.

While useful, AI is seen as a double-edged sword within the context of sustainability. The latest variants, which can analyse and generate information from new and existing data – otherwise known as “generative AI” – hold much potential to mitigate climate change. But the technology, which relies heavily on data centres to process mammoth amounts of information, also has a high energy and carbon footprint, which could skyrocket if left unattended.

Google wants to both capture the green opportunities that thinking machines afford and reduce the resultant emissions by investing in clean energy and efficient hardware. Key Asian financiers are getting on board, but also want to see the company addressing technology inequality in the fast growing region.

AI for climate action

Specialised research supercomputers have been conducting climate modelling and Earth observation for decades. Modern AI innovations promise to not only boost future efforts, but also democratise research and results – ideally to anyone with an internet-connected laptop or cellphone.

Case studies are already starting to emerge. The Gujarat Mahila Housing Sewa Trust, a non-profit in India, is using AI to predict flooding risks and affected zones in a small town, an effort already helping thousands of women farmers avoid hazards, shared Naina Batra, chief executive of the Asian Venture Philanthropy Network (AVPN).

Google panel

Batra was speaking at the APAC Digital Transformation Summit held in Google’s Singapore headquarters in June. She spoke alongside Kate Brandt, Google chief sustainability officer, and Stephanie Hung, Asian Development Bank’s director general of the information technology department. Image: Liang Lei/ Eco-Business.

Elsewhere in India, AI is providing farmers insights into water efficiency and managing groundwater levels. Over in Southeast Asia, the Southeast Asia Climate and Nature-based Solutions (SCeNe) Coalition, a group of non-profits promoting nature-based solutions in Southeast Asia, is using Google infrastructure to map nature-based carbon projects in the region that deliver co-benefits to local communities and biodiversity.

“When you hear about that 43 per cent reduction in emissions we need to achieve by 2030, [you know] we have a lot of work left to do,” said panelist Kate Brandt, Google’s sustainability chief, referring to the 1.5-degree Celsius limit set by the UN-backed Intergovernmental Panel on Climate Change to keep global warming within safe limits.

“How can we bring all the solutions to the table to accelerate action in this decisive decade? AI is a really powerful tool for enabling that,” Brandt added. Google completed the construction of its fourth data centre in the city-state this month, raising its investment in Singapore to US$5 billion and expanding its AI capabilities in the region.

Global energy-related emissions rose by 1.1 per cent in 2023, according to a report by the International Energy Agency (IEA) released that year. While the figure is a slight drop from the 1.3 per cent rise seen in 2022, the world is far off course to meet 2030 emissions targets. Proven AI technologies, if scaled up, could help the planet achieve 20 per cent of that goal, according to a joint report by Google and consultancy Boston Consulting Group in November 2023.

Key benefits include better data monitoring, greater renewables integration into power grids, and the enabling of carbon removal projects, the study said. Brandt added that the ability of AI systems to optimise processes can help to save fuel and lower emissions – for instance by finding ways to synchronise traffic lights to avoid unnecessary red-light stops, currently trialed in countries such as Indonesia and India.

In Asia Pacific, adapting to climate change is also worrying policymakers, as the bulk of its four billion people live in developing countries with limited ability to withstand large climate shocks.

Less than 5 per cent of adaptation technology comes to the Asia Pacific, Batra said. Many people remain victims of energy poverty and need help with power access, she added. AVPN is part of a US$8 million series of seed funding initiatives with the Asian Development Bank and Google’s philanthropic arm Google.org to scale up technology-led sustainability solutions in the region – such as better weather forecasting and disaster response.

Decarbonising the cloud

Tech companies’ efforts to put AI at people’s fingertips, however, come with potentially heavy carbon consequences. More data centres are needed to provide processing power and information storage space, and huge amounts of electricity are needed to run such facilities. Complex AI tools that can help to solve climate change may require extensive “training” before they can be fielded, while extensive Internet transmission infrastructure is needed to make software services available worldwide – both potentially adding to carbon emissions.

Data centres and transmission networks currently contribute to 0.6 per cent of total greenhouse gas emissions, according to the IEA in 2023, with cloud and hyperscale data centres representing only a part of that. Estimated global data centre electricity consumption in 2022 was 240-340 TWh, or around 1 to 1.3 per cent of global final electricity demand.

That said, the carbon footprint of running such infrastructure in Asia is likely higher, given the region’s heavy use of fossil fuels for electricity generation. Data centres in the balmy tropics will also require more power for cooling.

“The carbon footprint of technology and AI are legitimate fears. Every country, at least in this part of the world, is very interested in attracting more and more data centres to be set up,” said Batra. Civil society needs to be “asking critical questions” about the carbon footprint of data centre investments, she added.

Brandt said AI’s environmental footprint can be managed by improving the energy efficiency of data centre infrastructure, procuring more carbon-free energy for data centres, and using better training models in developing software tools. The company believes that it has accumulated significant experience in training AI models that are less compute-intensive without compromising on quality or speed.

As a result, Google’s data centres are three times as energy efficient today compared to five years ago, while its latest AI training tools have emissions footprints a thousand times lower, Brandt noted. Google is also aiming to run on “24/7” clean energy by 2030, meaning every hour of electricity consumption is matched with carbon-free electricity sources, on every grid where they operate.

Still, the challenges are stacked for Asia. Associate professor Lee Poh Seng, who heads the Energy Studies Institute at the National University of Singapore, said developing countries in the region may face financial and logistical barriers in implementing sustainable technologies at scale.

More research is needed to make low-carbon technologies affordable, while policymakers need to look into harmonising policies and incentivising businesses to adopt greener practices – such as trimming unnecessary computing and procuring equipment that emits fewer emissions, Lee added.

Sustainability concerns previously led Singapore to implement a four-year moratorium on data centre expansion from 2019 to 2022. But the ban has since been lifted and stricter environmental requirements are now in place for new developments. Studies are also underway on how data centres could run with less cooling without compromising their performance.

“The growth of AI is a striking example of the opportunities data centres enable us to capture,” said Dr Janil Puthucheary, Singapore’s senior minister of state for communications and information, at Google’s latest data centre launch this month, cautioning that new developments will face “energy and carbon constraints”.

“We are determined to turn these constraints into opportunity – an opportunity to innovate, so that we capture value from the growth of data centres while meeting our international climate commitments,” Puthucheary said.

Like this content? Join our growing community.

Your support helps to strengthen independent journalism, which is critically needed to guide business and policy development for positive impact. Unlock unlimited access to our content and members-only perks.

最多人阅读

专题活动

Publish your event
leaf background pattern

改革创新,实现可持续性 加入Ecosystem →