As artificial intelligence adoption accelerates across the Asia Pacific region, data centres face unprecedented demands. Legacy infrastructures, originally designed for less intensive computing tasks, are increasingly inadequate to support the soaring energy consumption and cooling requirements of contemporary AI workloads. Forecasts suggest that by 2030, GPU-centric operations could drive rack power densities close to 1 megawatt, rendering incremental facility upgrades insufficient. Consequently, the industry is pivoting towards purpose-built “AI factory” data centres engineered specifically for these challenges.
Rapid Expansion of AI Data Centre Market in Asia Pacific
The AI data centre sector is on track to expand dramatically, with market valuations expected to leap from $236 billion in 2025 to nearly $934 billion by 2030. This surge is fueled by AI integration across critical industries such as finance, healthcare, and manufacturing, all of which demand high-performance computing environments powered by dense GPU arrays. These setups require significantly enhanced power and cooling infrastructures compared to traditional server farms.
In Asia Pacific, this growth is further propelled by robust government initiatives promoting digital transformation, widespread 5G deployment, and the rapid adoption of cloud-native and generative AI technologies. These factors collectively drive compute requirements at an unprecedented pace, challenging existing data centre capabilities.
Innovative Infrastructure Strategies for Scalable AI Support
Paul Churchill, Vice President of Vertiv Asia, emphasizes that addressing these demands necessitates more than just expanding physical space. “Operators must transition from fragmented upgrades to comprehensive, AI-optimized infrastructure strategies that integrate high-capacity power delivery, cutting-edge thermal management, and scalable, modular designs,” he explains. This holistic approach ensures data centres remain adaptable and sustainable amid evolving AI workloads.
Overcoming Power and Cooling Constraints
With rack power densities projected to climb from 40 kW today to as high as 250 kW by 2030, traditional air cooling methods are proving inadequate. To meet these challenges, hybrid cooling solutions combining direct-to-chip liquid cooling with advanced air cooling are being developed. These systems dynamically adjust to workload fluctuations, enhancing energy efficiency and maintaining operational reliability. Churchill notes, “Our coolant distribution units facilitate direct liquid cooling while ensuring serviceability in ultra-dense environments.”
Power delivery systems are also evolving to accommodate the volatile nature of AI workloads. Vertiv is enhancing rack power distribution units and busway architectures to support higher voltages and improve load balancing. Intelligent monitoring tools enable operators to optimize power usage, minimize waste, and maximize uptime-critical in regions like Southeast Asia where grid stability can be inconsistent.
Architectural Transformation: Designing Data Centres for AI
The emergence of liquid-cooled GPU clusters and 1 MW rack configurations, as seen in initiatives by AMD and hyperscale providers such as Microsoft, Google, and Meta, signals a fundamental shift in data centre design philosophy. Rather than retrofitting existing facilities, new data centres are being purpose-built to accommodate AI’s unique demands.
“Future data centre architectures will be hybrid, centered around liquid cooling infrastructure,” Churchill states. This entails reimagined floor plans, sophisticated coolant distribution networks, and integrated power systems designed from the ground up. Such comprehensive integration-from chip-level cooling to grid-level power management-is essential for Asia Pacific’s rapidly expanding hyperscale campuses to meet both performance and sustainability targets.
Transitioning to AI Factory Data Centres: A Phased Approach
By 2030, Asia Pacific is projected to surpass the United States in data centre capacity, reaching nearly 24 gigawatts of commissioned power. To manage this scale, enterprises are shifting from incremental upgrades to fully integrated AI factory data centres.
Churchill outlines a three-step transition: first, integrated planning that unifies power, cooling, and IT management systems to streamline deployment and scalability; second, adoption of modular, prefabricated components that enable phased capacity expansion with minimal disruption; and third, embedding sustainability through technologies like lithium-ion energy storage, grid-interactive uninterruptible power supplies (UPS), and high-voltage distribution systems to enhance efficiency and resilience.
Direct Current (DC) Power: Enhancing Efficiency and Sustainability
Vertiv’s recent launch of the PowerDirect Rack-a DC power shelf tailored for AI and high-performance computing-highlights the growing relevance of DC power in data centres. By reducing energy losses associated with multiple AC-DC conversion steps, DC power systems improve overall efficiency and align seamlessly with renewable energy sources and battery storage solutions increasingly deployed across Asia Pacific.
This approach is particularly advantageous in energy-constrained markets such as Vietnam and the Philippines, where flexible and efficient power management is vital. Churchill remarks, “DC power is more than an efficiency measure; it’s a strategic enabler for sustainable, scalable AI infrastructure.”
Embedding Sustainability at the Core of AI Data Centres
As AI workloads drive up energy consumption, data centre operators face mounting regulatory pressures and grid limitations, especially in Southeast Asia where power reliability and costs vary significantly. Vertiv collaborates with operators to integrate alternative energy solutions including lithium-ion batteries, hybrid power configurations, and microgrids, reducing grid dependency and enhancing resilience.
Interest is also growing in solar-powered UPS systems and advanced energy storage technologies that help balance demand and control operational expenses. Cooling innovations, such as hybrid liquid cooling, contribute to reducing both energy and water consumption compared to legacy methods. Churchill emphasizes, “Our mission is to deliver infrastructure that meets rigorous performance standards while advancing environmental, social, and governance (ESG) objectives. We work closely with partners to ensure AI-driven growth in the region is responsible and sustainable.”
Modular Data Centres: Addressing Regional Challenges with Agility
Emerging economies in Asia Pacific often contend with limited land availability, unstable power grids, and workforce skill shortages. Modular and prefabricated data centre solutions offer a practical response to these constraints, enabling faster deployment-up to 50% quicker-while enhancing energy efficiency and scalability.
These modular systems allow operators to incrementally increase capacity in line with unpredictable AI workload growth, minimizing upfront capital expenditure and operational risk. By combining compact footprints with energy-conscious designs, modular data centres provide a strategic advantage as the region’s digital economy expands rapidly.
Preparing for the Next Wave of AI-Driven Computing
The AI revolution is fundamentally transforming data centre construction and management across Asia Pacific. As computational demands intensify and sustainability imperatives grow, reliance on outdated infrastructure is no longer viable. The shift toward AI factory data centres-featuring advanced cooling technologies, DC power integration, and modular architectures-reflects a proactive strategy to meet the region’s future computing needs efficiently and responsibly.
