Headlines about artificial intelligence in Asia Pacific tend to follow the biggest numbers: larger and more powerful data centers, bigger investment commitments, and ever-higher forecasts for what the region will spend pursuing its AI ambitions.
Yet beneath these headline figures lies a more fundamental question: Can the region generate and deliver enough power to support those ambitions?
As AI scales, power—more than capital or chips—is rapidly becoming a key constraint, changing the definition of infrastructure leadership.
Demand is not the question. IDC forecasts data center power capacity across Asia Pacific, excluding Japan, to reach approximately 142,600 megawatts by 2029, while Gartner expects global data center electricity demand to double by 2030.
The bottleneck is no longer demand but supply. With grid connection wait times exceeding four years in major markets, access to reliable power is rapidly becoming a strategic differentiator. For enterprise leaders, this turns what was once a facilities and operations question into one of corporate strategy.
The move from more compute to better results
For the past decade, the response when AI demanded greater scale was simply to add more: more accelerators, more racks, and more capacity.
When faced with a power constraint, however, that approach works against an organization. The more important question becomes how much value each unit of power can produce, rather than how much computing capacity can be deployed.
In other words, performance per watt is emerging as one of the most important metrics for AI infrastructure. Efficiency stops being a footnote in sustainability reports and becomes a practical factor in how far an organization can scale with the resources it has.
This requires a tangible shift in mindset, with real commercial implications. In the face of rising energy costs, electricity is already the largest operating cost for many data centers. The efficiency of every server, rack, and cooling choice therefore belongs in the boardroom rather than only in the server room.
Organizations that can extract more useful work from every watt will have greater room to grow. Those that treat efficiency as an afterthought may find their options narrowing sooner than expected.
Policymakers are already drawing the line. In late 2025, Singapore’s Economic Development Board and Infocomm Media Development Authority opened a second Data Centre Call for Application, or DC-CFA2, directing at least 200 megawatts of new capacity to operators that can meet some of the region’s most demanding standards for energy overhead, equipment efficiency, and green power.
Meanwhile, Malaysia has tied data center tax incentives to energy-efficiency and emissions targets as part of its wider push for 70 percent renewable energy and net-zero emissions by 2050.
Lenovo has also set a target of reaching net-zero greenhouse gas emissions by 2050, validated by the Science Based Targets initiative. This ties the efficiency gains described here to a measurable commitment rather than a general ambition.
Design for the whole life of the workload
One of the biggest design mistakes leaders can make is optimizing only for the workload in front of them.
Much of the public conversation around AI focuses on training large models, which is intensive but occurs in bursts. As organizations move from experimentation to deployment, the inference workload becomes continuous: a model answering a customer, checking a payment, or guiding a process on a factory line.
The power draw per task may be smaller, but it is constant and accumulates day after day. At a hypothetical cost of one cent per prompt, a service with 10 million users making 10 requests each day would spend approximately $1 million daily on electricity and computing.
JLL projects that inference will overtake training as the dominant data center AI workload around 2027, which means this operating cost will continue to rise as AI deployment matures.
The answer is to treat architecture as a strategic decision and design for the entire life of a workload.
In practice, this means choosing infrastructure according to performance per watt rather than headline speed, sizing systems for sustained real-world demand rather than the busiest possible day, and matching each workload to the environment in which it performs best.
Place workloads where they belong
Where a workload runs is one of the most direct levers leaders have over cost, performance, and energy use. This points toward a hybrid approach.
Large training runs should reside where dense computing capacity and the required cooling are available. Inference, by contrast, often sits closer to where data is created and decisions are made, including at the edge.
This can reduce latency, bandwidth use, and power consumption. Keeping workloads closer to their data can also help organizations address the data-sovereignty and compliance requirements tightening across the region.
There is no single correct environment—only the right environment for each workload.
Hybrid AI, spanning on-premises systems, the edge, and the public cloud, is becoming the default enterprise architecture for that reason. Treating workload placement as a deliberate design choice, rather than a default decision, helps keep both energy costs and response times in check as AI scales.
Efficiency is a team sport
No single component can deliver this kind of efficiency on its own.
The gains come when silicon, system design, software, and cooling are engineered to work together, with each element easing the limitations of the others. Meaningful progress depends on collaboration across the technology stack among chipmakers, infrastructure vendors, and the organizations deploying AI.
Cooling has moved from the back of the data hall to the center of infrastructure strategy. Enterprises want more of their available power and budget to support the use of AI to grow the business rather than simply remove heat.
As AI racks run hotter, air cooling begins to reach its limits. Keeping dense racks cool with air requires more fans and greater reliance on chillers. That supporting equipment adds to energy costs beyond the power consumed by computing equipment itself.
Reducing cooling overhead is therefore one of the fastest ways to improve data center efficiency.
Neptune warm-water cooling illustrates what this can look like in practice. By circulating warm water directly to the hottest components, it reduces the need for the power-intensive fans and chillers on which air-cooled data halls depend.
Lenovo reports that the technology can reduce data center power consumption by up to 40 percent compared with similar air-cooled systems. Those savings come from the cooling side of the equation—from fewer fans and no chillers—rather than from servers drawing less power to perform the same work.
In some systems, this can bring power usage effectiveness as low as 1.1, below the levels typical of conventional designs. Such systems are already in use among meteorological agencies, universities, and research institutions across Asia Pacific.
When the entire technology stack is designed for efficiency, the savings add up.
Why Asia Pacific can lead
There is an environmental dividend to be gained from this shift. Lower energy use means lower emissions, helping organizations respond to the disclosure expectations tightening across the region.
The more immediate reason efficiency is rising on the corporate agenda, however, is simpler: Where power is constrained, efficiency is what keeps growth within reach, regardless of the local energy mix.
The commercial and environmental cases point in the same direction.
Asia Pacific also has one critical advantage: Much of the region’s AI-capable capacity is being built now, allowing efficiency to be designed into infrastructure from the beginning.
The leaders that pull ahead will be those that spend well rather than simply spend more. They will treat infrastructure as a series of decisions about efficiency and workload placement rather than as a race to add capacity.
In the next phase of AI adoption, competitive advantage will not come from deploying the most infrastructure. It will come from extracting the greatest value from every watt.
Where power, more than ambition, is the scarce input, organizations that make the most of every watt will be best positioned to scale.
Kumar Mitra is the Executive Director, Central Asia Pacific and Australia & New Zealand, Infrastructure Solutions Group, at Lenovo.
As Executive Director for CAP (Central Asia Pacific) and ANZ (Australia & New Zealand) of Lenovo Infrastructure Group, Kumar is responsible for building customer confidence and accelerating growth across 9 key markets in CAP, consisting of Singapore, Hong Kong, Malaysia, Thailand, Indonesia, Vietnam, Taiwan the Philippines, and Australia & New Zealand.
Kumar’s areas of expertise include establishing strategic growth across channel, direct, and hybrid business development. He joins Lenovo ISG with over 26 years of experience in business and technology optimization. Kumar held previous director-level positions at Nutanix and Dell where he played a key role in transforming the business with partner growth and customer acquisition.
A strong believer in developing and maintaining a highly collaborative culture, Kumar’s leadership style focuses on the 4As – Attention to detail, Anti-fragility, Absolute integrity and Ambitious. He has also obtained several awards including WW GTM Top Contributor in Nutanix, and Inspiring Leader of the Year at Dell, APJ.
Kumar holds a Bachelor’s Degree in Electronics from Doctor Babasaheb Ambedkar Marathwada University.
TNGlobal INSIDER publishes contributions relevant to entrepreneurship and innovation. You may submit your own original or published contributions subject to editorial discretion.
Featured image: Logan Voss on Unsplash
Empowering Singapore’s SMBs to build an AI-ready IT foundation for growth