The growing wave of AI headlines may suggest that the Asia Pacific’s power infrastructure is nearing a breaking point. Yet, the numbers paint a far more balanced picture. The majority of data center energy consumption today, and for the foreseeable future, does not come from AI. It comes from everything else: the conventional enterprise servers, colocation workloads, and cloud infrastructure that keep businesses running every day. Mainstream IT, it turns out, is the region’s biggest digital energy consumer, and it is not a title AI is close to claiming.
AI’s expanding energy demands warrant legitimate concern. But concentrating solely on AI risks overlooks the much larger opportunity to improve efficiency across the entire IT landscape. In a region witnessing rapid data center expansion, particularly across Singapore, Australia, Japan, and India, the challenge of reconciling digital acceleration with sustainability goals is becoming increasingly critical, making this a risk the industry cannot afford to overlook. Despite this, sustainability debates continue to default to AI and mainstream compute, leaving the region’s real energy-intensive challenges insufficiently addressed.
Beyond the headlines: Where the energy is actually going
The data on this is unambiguous. Global research bodies estimate that AI accounted for just 10 to 15 percent of worldwide data center energy consumption in 2024, with general-purpose IT making up the remaining 85 to 90 percent. There is also reason to think that even those AI projections may be on the high side, given the uncertainty around whether all planned large-scale data center builds will actually proceed.
Across the Asia Pacific, this picture is particularly striking. In fact, as per Deloitte’s report – Powering Asia Pacific’s data center boom, the region is one of the fastest-growing data center markets in the world, with approximately $800 billion in data center investment expected across the region by 2030. Yet the vast majority of that installed base, and the energy it consumes, is running conventional workloads: ERP systems, financial platforms, government services, healthcare applications. These are not AI systems. They are the digital backbone of the regional economy, and they are consuming power at scale.
None of this diminishes the importance of addressing AI’s energy trajectory. AI training workloads introduce volatile demand patterns that create real challenges for grid stability, and the pace of AI infrastructure growth demands proactive planning. The point is simply that conventional IT is not standing still either, and for as long as sustainability strategies focus only on AI, the world’s real digital workhorses will continue consuming far more electricity than the headline-grabbing systems, largely unaddressed.
The efficiency opportunity hiding in plain sight
The good news is that the tools to address mainstream IT’s energy footprint are already well understood. Across data, software, infrastructure, and facilities, there is a proven playbook, and it does not require waiting for the next hardware cycle. Enterprises can achieve meaningful reductions by rationalising storage strategies, decommissioning or modernising inefficient legacy applications, refreshing ageing server fleets with more efficient systems, and improving workload placement to drive higher utilisation rates.
On the facilities side, better airflow management, direct liquid cooling, and heat recovery technologies are all increasingly accessible and deliver efficiency gains across AI and conventional workloads alike. Advanced cooling alone has the potential to reduce total data center energy consumption by as much as 5–15 percent, as per PwC’s report – Powering possibility: Closing the clean energy gap for Asia Pacific data centres. In land-scarce, high-density markets like Singapore and Hong Kong, where constraints on physical footprint make operational efficiency a baseline requirement, these interventions are particularly consequential. The same pressure, if for different reasons, applies across the broader region: Japan and Australia, both committed to net-zero by 2050, face tightening regulatory expectations on data center sustainability, while fast-growing markets such as India, Malaysia, and Indonesia are navigating the challenge of scaling infrastructure without compounding already strained energy and water resources.
The focus on mainstream IT efficiency and AI efficiency must go hand in hand, not compete for attention. The techniques largely overlap, and the organisations that approach both in parallel will be better placed on cost, carbon, and operational resilience. Taking a whole-of-IT view is not a compromise; it is the more rigorous approach.
Across the Asia Pacific, there is a growing recognition that IT efficiency is not just an environmental question but a business-critical one. Regulators in Singapore, Australia, and Japan have all signalled heightened expectations around data center sustainability. Institutional investors are applying greater scrutiny to digital infrastructure operators’ energy performance. Enterprise customers are increasingly factoring IT carbon footprint into procurement decisions. For organisations in this region, demonstrating responsible energy stewardship is becoming a baseline commercial requirement. Not a differentiator, but a licence to operate.
The organizations that will lead the next era of IT sustainability are those that resist the pull of the headlines and take stock of the full picture, optimizing not just the AI systems attracting attention today, but the entire computing estate that has been quietly consuming energy all along.
Shankar Raghavan is Senior Director & General Manager, Advisory & Professional Services, HPE Asia Pacific.
TNGlobal INSIDER publishes contributions relevant to entrepreneurship and innovation. You may submit your own original or published contributions subject to editorial discretion.
Featured image: Adrien on Unsplash
Southeast Asia’s $30 billion data center boom is racing into a power wall