The smart artificial intelligence design, associated with green data infrastructure, in Southeast Asia, can help meet its digital ambitions without compromising the goals of energy transmission. Not only does the area need more green data infrastructure – it also needs artificial intelligence applications that do more more.

The digital economy in Southeast Asia flourishes. With rapid growth in e-commerce, Fintech and AI services, the region is witnessing an increase in electricity demand-especially from databases. These facilities operate around the clock throughout the week and require intense cooling, putting large loads around the clock on national energy networks.

Globally, the consumption of data centers 415 TwH From electricity in 2024 – more than IndonesiaThe entire national consumption. By 2030, electricity is expected to be used Override That Japan today. While much of the expansion of the global data center occurs in the United States, China, Europe and Southeast Asia-on which the Singapore Data Center-Malisia-Sanean is based More than twice By 2030.

Country estimates Highlighting the challenge scale. In Malaysia, the demand for electricity from databases may rise seven times by 2030, as it reached nearly 30 percent of national consumption. In Indonesia, it is expected to require almost quadruple, while in the Philippines, it may rise more than eighteen times. The increasing risks of the data center also risk competition with residential and societal needs of electricity and water – especially in areas with restricted and limited networks Water supply Raise social concerns or wider shares.

If this increasing demand is often met by heavy fossil networks, then it is that Risk Slowness – or even exit – the transmission of clean energy in the region. As of 2022, fossil fuels, led by coal, are still supplied More than 70 percent From Electricity in Southeast Asia, despite the continuous expansion of renewable energy sources.

The main part of the solution is to improve devices, especially through the development of “development”Green data centers“These facilities adopt advanced technologies such as highly efficient cooling systems, waste recycling, diverting the work burden into peak hours, and the integration of renewable energy sources. With these improvements, data centers can become more effective in the energy field, decisively, as cranes to accelerate clean energy.

Southeast Asian countries are already moving in this direction. Singapore 2024 Green Data Center Map It sets first -class energy efficiency standards and provides incentives to use renewable energy. Malaysia is preparing to launch a sustainable framework for the data center By late 2025.

These initiatives determine an important progress in improving the devices layer for digital infrastructure. yet Fears It remains whether the equipment improvements alone will be sufficient. Meanwhile, other strong wheels remain unused, especially in the software layer. One of the opportunities lies in the most intelligent artificial intelligence design and owners – building applications for the same results with a lower accounting heavy lifting, which reduces the demand for infrastructure and energy.

In practice, this can be achieved by publishing Small models of artificial intelligence Instead of the sprawling parties for general purposes, using Smaller data sets but the highest quality In typical training, presentation Model pressure techniques Like pruning and quantity to reduce arithmetic pregnancy and adoption More efficient algorithms For each of the training and inference.

These measures have great potential to improve software efficiency and reduce energy use. For example, at Google I mentioned Its Gemini model, which combines the most efficient software structure and algorithms with equipment improvement, consumes much less energy than many previous general estimates.

To pick up these potentials completely, continuous research support is needed, including more focus on Amnesty International Green OrganizationAlong with broader artificial intelligence initiatives and applications for the sector. But the most urgent challenge is not technical knowledge; It creates an environment to enable the right.

For years, artificial intelligence developers – from the foundation model engineers to applications – have been rewarded for accuracy, speed and features, not energy efficiency. that it Starting change Since the high account costs and the distinctive symbol compels efficiency in the conversation, most efforts remain dedicated. Without a clear political signal to include efficiency in developing artificial intelligence application, progress can stop, and intense energy programs can prevail if energy costs decrease or prioritize.

This is where governments and companies can work together. Instead of organizing the design of artificial intelligence directly, policymakers can enhance an enabled environment by enhancing reporting criteria for energy use of artificial intelligence applications and supporting voluntary efficiency standards. Companies, in turn, can cooperate by sharing data, experimenting with light weight applications, and displaying best practices in improving algorithms. Public agencies should also consider setting the priorities of the basic social needs on the estimated use, ensuring that the networks continue to serve the interests of the broader society as the demand for artificial intelligence grows.

The path is not clear. Many artificial intelligence applications, such as news summarizing tools, depend on the third -party systems that are often unknown for applications for applications. This ostrich makes it difficult to set responsibility and measure the impact and reward.

To move in this challenge, the first vital step is to realize that software efficiency is crucial like devices. Such recognition is the starting point necessary for the deeper discussion between the stakeholders concerned about the best ways to align digital growth with environmental sustainability.

Dr. Moy Yang is a major policy analyst at Ember. He also holds positions as a colleague of an old non -resident policy at the Asia Australia Association and an assistant colleague at the Chinese China Institute of China at the University of Sydney.

Xiwei (Sherry) Xu is a major research scientist in Csiro’s Data61.

The smart artificial intelligence design, associated with green data infrastructure, in Southeast Asia, can help meet its digital ambitions without compromising the goals of energy transmission. Not only does the area need more green data infrastructure – it also needs artificial intelligence applications that do more more.

The digital economy in Southeast Asia flourishes. With rapid growth in e-commerce, Fintech and AI services, the region is witnessing an increase in electricity demand-especially from databases. These facilities operate around the clock throughout the week and require intense cooling, putting large loads around the clock on national energy networks.

Globally, the consumption of data centers 415 TwH From electricity in 2024 – more than IndonesiaThe entire national consumption. By 2030, electricity is expected to be used Override That Japan today. While much of the expansion of the global data center occurs in the United States, China, Europe and Southeast Asia-on which the Singapore Data Center-Malisia-Sanean is based More than twice By 2030.

Country estimates Highlighting the challenge scale. In Malaysia, the demand for electricity from databases may rise seven times by 2030, as it reached nearly 30 percent of national consumption. In Indonesia, it is expected to require almost quadruple, while in the Philippines, it may rise more than eighteen times. The increasing risks of the data center also risk competition with residential and societal needs of electricity and water – especially in areas with restricted and limited networks Water supply Raise social concerns or wider shares.

If this increasing demand is often met by heavy fossil networks, then it is that Risk Slowness – or even exit – the transmission of clean energy in the region. As of 2022, fossil fuels, led by coal, are still supplied More than 70 percent From Electricity in Southeast Asia, despite the continuous expansion of renewable energy sources.

The main part of the solution is to improve devices, especially through the development of “development”Green data centers“These facilities adopt advanced technologies such as highly efficient cooling systems, waste recycling, diverting the work burden into peak hours, and the integration of renewable energy sources. With these improvements, data centers can become more effective in the energy field, decisively, as cranes to accelerate clean energy.

Southeast Asian countries are already moving in this direction. Singapore 2024 Green Data Center Map It sets first -class energy efficiency standards and provides incentives to use renewable energy. Malaysia is preparing to launch a sustainable framework for the data center By late 2025.

These initiatives determine an important progress in improving the devices layer for digital infrastructure. yet Fears It remains whether the equipment improvements alone will be sufficient. Meanwhile, other strong wheels remain unused, especially in the software layer. One of the opportunities lies in the most intelligent artificial intelligence design and owners – building applications for the same results with a lower accounting heavy lifting, which reduces the demand for infrastructure and energy.

In practice, this can be achieved by publishing Small models of artificial intelligence Instead of the sprawling parties for general purposes, using Smaller data sets but the highest quality In typical training, presentation Model pressure techniques Like pruning and quantity to reduce arithmetic pregnancy and adoption More efficient algorithms For each of the training and inference.

These measures have great potential to improve software efficiency and reduce energy use. For example, at Google I mentioned Its Gemini model, which combines the most efficient software structure and algorithms with equipment improvement, consumes much less energy than many previous general estimates.

To pick up these potentials completely, continuous research support is needed, including more focus on Amnesty International Green OrganizationAlong with broader artificial intelligence initiatives and applications for the sector. But the most urgent challenge is not technical knowledge; It creates an environment to enable the right.

For years, artificial intelligence developers – from the foundation model engineers to applications – have been rewarded for accuracy, speed and features, not energy efficiency. that it Starting change Since the high account costs and the distinctive symbol compels efficiency in the conversation, most efforts remain dedicated. Without a clear political signal to include efficiency in developing artificial intelligence application, progress can stop, and intense energy programs can prevail if energy costs decrease or prioritize.

This is where governments and companies can work together. Instead of organizing the design of artificial intelligence directly, policymakers can enhance an enabled environment by enhancing reporting criteria for energy use of artificial intelligence applications and supporting voluntary efficiency standards. Companies, in turn, can cooperate by sharing data, experimenting with light weight applications, and displaying best practices in improving algorithms. Public agencies should also consider setting the priorities of the basic social needs on the estimated use, ensuring that the networks continue to serve the interests of the broader society as the demand for artificial intelligence grows.

The path is not clear. Many artificial intelligence applications, such as news summarizing tools, depend on the third -party systems that are often unknown for applications for applications. This ostrich makes it difficult to set responsibility and measure the impact and reward.

To move in this challenge, the first vital step is to realize that software efficiency is crucial like devices. Such recognition is the starting point necessary for the deeper discussion between the stakeholders concerned about the best ways to align digital growth with environmental sustainability.

Dr. Moy Yang is a major policy analyst at Ember. He also holds positions as a colleague of an old non -resident policy at the Asia Australia Association and an assistant colleague at the Chinese China Institute of China at the University of Sydney.

Xiwei (Sherry) Xu is a major research scientist in Csiro’s Data61.