The data centre construction market in Southeast Asia (SEA) is projected to grow at a compound annual growth rate (CAGR) of 11.18 per cent, reaching $5.29 billion between 2023 and 2029. This expansion is driven by increasing data consumption, supportive government policies and technological progress. To break this down, studies indicate that approximately 125,000 new internet users are added daily to SEA and monthly data usage in the region is expected to increase from 9.2 GB in 2020 to 28.9 GB per user by 2025. State initiatives such as Malaysia’s JENDELA broadband project (2020), Singapore’s Green Data Centre Roadmap (2024) and Vietnam’s Cybersecurity Law (2014) have laid the foundation for this industry to thrive. In terms of technology, the conjunction of artificial intelligence (AI) and data centres has become one of the most significant trends shaping the region’s technological landscape. This rapid digital expansion sets the stage for a deeper look at how AI is reshaping the very foundation of modern data centres.
Although conventional data centres typically have many of the same mechanisms as an AI-optimised data centre, the latter has more advanced computing tasks (such as machine learning (ML), deep learning, and generative AI), network and storage architectures, energy efficiency and cooling capabilities. For example, traditional data centres are more likely to be constructed for and contain central processing units (CPUs) whereas AI data centres need high-performance graphics processing units (GPUs) and tensor processing units (TPUs) that expedite tasks like data processing and model training. Additionally, AI is enhancing facility management at data centres by predicting hardware outages, automating cooling systems and intelligently allotting resources to maximise energy efficiency.
A look at the components of an AI-ready data centre, investment trends and the energy needs of the region…
Reimagining data centres
Data centre traffic is growing explosively. Technologies like reconfigurable data centre networks (RDCNs) are gaining traction in AI-ready data centres. As optical switching technology has advanced, RDCNs have become a creative and alluring substitute for conventional static topology data centre network architectures. Compared to the fixed nature of state-of-the-art data centre network topologies, these dynamic topology-based data centre networks constitute a paradigm change. These networks use optical switches, software-defined networking (SDN), and disaggregated hardware to adapt dynamically to traffic demands, enhancing performance and resource utilisation.
In parallel, another transformative technology, digital twins, is redefining operational intelligence in AI data centres. In AI data centres, this means creating a virtual replica of critical physical systems – including servers, power and cooling infrastructure, networking components and even AI workloads. They can monitor hardware conditions, predict failures, reduce downtime and track parameters like power usage effectiveness (PUE) and water usage effectiveness (WUE). According to estimates, digital twins can help data centre operators lessen energy costs by up to 40 per cent without making any change to their hardware.
Investment activity
A Boston Consulting Group report titled “Accelerating Compute Needs Underpin Southeast Asia’s Rapid Data Center Growth”(October 2024) states that data centre capacity in SEA is expected to triple to reach between 5.2 GW and 6.5 GW by 2030. This surge will be fuelled by a remarkable tenfold increase in demand for AI computing.
With AI-ready design becoming the new standard, capital is pouring in to support the next wave of high-performance infrastructure. According to Deloitte’s “Southeast Asia’s Data Centres and AI Infrastructure Imperative” report (March 2025), setting up a 100 MW AI-ready data centre will cost at least $1 billion. On the brighter side, data centres promise stable cash flows. One MW of AI-ready power can generate $1.5 to $2 million annually and probably comes with a 10-year commitment or more.
An industry estimate predicts that investment in AI and hyperscale data centres in SEA is likely to grow from $10.23 billion in 2023 to $17.73 billion by 2029.
These projections have already spurred a wave of high-profile investments across SEA over the past year. For instance, in May 2024, ST Telemedia Global Data Centres (STT GDC) announced AI readiness in its data centres across SEA – Singapore, Malaysia, Thailand, Indonesia, the Philippines and Vietnam. In May 2024, Microsoft promised $1.7 billion to build AI infrastructure in Indonesia, while in July 2024 BDx Indonesia announced Phase 1 completion of its renewable-powered AI data centre park, scalable up to 500 MW. Similarly, in June 2024, TM and Nxera (the digital arm of Singtel) formed a joint venture to develop sustainable, hyperconnected AI-ready data centres in Malaysia. Further, in September 2024, Google announced its plans to invest $1 billion in Thailand to build a data centre and cloud region to meet growing cloud demand and support AI adoption in SEA. Microsoft, too, has indicated that Thailand would be the location for its lead regional data centre. Finally, in November 2024, Equinix announced its AI data centre in Singapore’s Tanjong Kling, which will be built with an initial investment of $260 million and is slated to open in Q1 2027.
Every watt and drop counts
Data centres, especially the AI-integrated ones, are massive energy guzzlers and need constant electricity supply with high levels of redundancy. Now ML algorithms are revolutionising data centre efficiency by intelligently predicting optimal cooling schedules and methods. These algorithms analyse variables such as local weather conditions, server traffic patterns and equipment usage trends to dynamically adjust cooling strategies. The result is a significant reduction in energy consumption and operational costs, contributing to a smaller carbon footprint. Moreover, AI technologies are increasingly being deployed to manage and integrate renewable energy sources – such as solar and wind – into the power mix, further enhancing the sustainability profile of data centre operations. In addition, other solutions like power purchase agreements and solar integration are also being explored to reduce the sector’s carbon footprint.
Another area of innovation is cooling technology as water usage is emerging as a critical issue. There is consensus within the industry that the way data centres are powered and cooled will influence the industry’s environmental footprint. With the industry moving towards hybrid approaches that combine air and liquid cooling, next-generation cooling solutions have begun to gain momentum. The major components of hybrid cooling include air cooling (vital for lower-density racks), direct-to-chip liquid cooling (targets heat-intensive elements like GPUs and CPUs), immersion cooling (designed for high-density deployments) and adiabatic cooling (crucial for water-scarce regions). The key benefits of hybrid cooling are lower PUE, reduced energy consumption and decreased operational expenditure.
Looking ahead, these innovations are laying the groundwork for the next phase of SEA’s digital evolution.
Outlook
As AI adoption continues to surge, the integration of AI and data centre infrastructure in SEA is poised to accelerate even further in the coming years. To meet the growing demands of AI-powered technologies, data centres across the region will need to evolve – embracing energy-efficient architectures, AI-optimised designs and strategic investments in high-performance computing hardware.
