For years, the global data centre industry has followed a predictable path: design, refine, and replicate at scale elsewhere. While that formula may work in a world where workloads are stable, regulations are standardised, and environmental conditions are often relatively predictable, a model that doesn’t really work in APAC.
The region isn’t simply growing faster. It is operating under conditions that expose the weaknesses of legacy infrastructure thinking and, in doing so, is forcing the industry to rethink how data centres are engineered, deployed, and evolved.
Why APAC Is forcing a rethink
APAC is not a single market. It is diverse in climates, non-standardised in regulatory regimes, and a variety of infrastructure maturity levels. The region necessitates a fresh engineering blueprint that accommodates these challenges and enhances resilience in its design.
Dealing with tropical humidities and temperature volatilities made worse with AI racks demanding up to 100 kW per cabinet creating a surge in heat densities, mean traditional cooling assumptions quickly collapse.
It can be difficult to design for AI operating in a temperate climate, but designing AI to function in tropical environments is even more challenging. These environmental factors push infrastructure to its limits, requiring a totally different mindset. APAC operators must engineer for both extremes simultaneously, without compromising uptime.
That pressure is driving more forward-looking thermal strategies, earlier planning for high-density deployments, and infrastructure that is adaptable rather than reactive.
Regulatory complexity in APAC adds yet another layer of infrastructure pressure with frameworks differing across all 15 major APAC economies. Regulations that relate to energy efficiency, sustainability requirements, and data sovereignty laws also vary across the region.
There is no one-size-fits-all deployment model that works. Infrastructure needs to be sufficiently standardised to sustain performance, modular enough to accommodate changes, and, crucially, locally aware to prevent costly errors.
APAC includes some of the world’s most advanced digital economies alongside rapidly developing markets. However, power reliability, grid stability, and specialised workforce experience and availability can vary greatly across the region.
That variability forces a new discipline in design, insisting that supply chains must be shortened, local engineering capability must be integrated early and redundancy must be intentional.
Why we are up against it
Japan
Japan's data centre market sits at the junction of unprecedented demand and tightening constraints.
According to Technology Magazine, AWS, Oracle and Microsoft have collectively invested more than US$26 billion in hyperscaler capital into Japanese AI infrastructure.
However, Wood Mackenzie reports that data centres are expected to account for 7 percent of the power load in this region by 2030; power that could potentially serve 15-18 million Japanese households.
With power consumption expected to triple to as much as 66 TWh by 2034, driving 60 percent of the country's total electricity demand growth, the power connections needed to support this growth could take up to a decade to complete.
This constraint manifests most acutely in Tokyo, where the gap between investment commitments and deliverable capacity is impossible to ignore.
Singapore
In 2019, the Singaporean government imposed a moratorium on new data centre construction due to resource constraints.
At that time, data centres were consuming about 7 percent of national electricity. The design layouts didn’t fit a real estate-restricted city, liquid-cooling systems were straining water resources, air-cooling approaches were inefficient in a tropical climate, and growth was unsustainable.
When the moratorium was lifted in 2022, there were more than 70 operational data centres in Singapore. This accounted for 60 percent of Southeast Asia’s total data centre capacity – a country that imports most of its electricity and produces little clean energy.
It became obvious that the AI-era demand could not be met by simply scaling legacy infrastructure. In June 2023, the Infocomm Media Development Authority of Singapore (IMDA) launched a new sustainability standard, establishing guidelines for the industry to safely raise the operating temperature of data centres in tropical climates and at higher humidity levels.
Leapfrogging legacy infrastructure
The pressures and challenges are not slowing the region down. In fact, they are accelerating innovation. Because operators cannot rely on legacy assumptions, they are bypassing them entirely. Instead of retrofitting outdated models, APAC is demanding:
- Higher-density-ready infrastructure from day one
- Modular scalability that can evolve with workload demand
- Supply chain agility
- Embedded engineering partnerships instead of distant vendor relationships
By anticipating AI acceleration, regulatory complexity, and unstable environments as ordinary operating circumstances, this leapfrogging impact is producing an alternative, proactive blueprint. Global markets are beginning to study this approach closely, because the constraints APAC faces today will be emerging elsewhere tomorrow.
Success in APAC does not come from exporting pre-packaged global templates. It comes from embedding expertise, starting with the physical and operational truth of the location - not with a design that worked elsewhere.
- Midge Pan, General Manager, APAC, Subzero Engineering.
Maintaining consistent global standards while investing in regional manufacturing, local engineering talent, and partner integration makes it possible to deliver scalable solutions without sacrificing reliability.
The operational reality of each site locality means assessing and understanding humidity loads before finalising airflow containment strategy.
Designing for Singapore's humidity differs vastly from Sydney's temperate climate. In tropical environments, moisture control isn’t a secondary consideration; it directly influences material selection, corrosion resistance, cable management durability, and long-term maintenance cycles.
It means anticipating AI-driven heat density before racks are installed, not after. AI clusters introduce sustained, concentrated thermal output that challenges traditional aisle containment assumptions. In many APAC markets, facilities are being built in dense environments where expansion space is limited, so infrastructure must be designed for higher density from day one.
Licensing processes, sustainability requirements, and data sovereignty rules vary not only by nation but also frequently by locality as well. Instead of navigating the approval landscape reactively, teams on the ground with local insight and knowledge can do it proactively.
Developing local engineering talent is essential. Potential risks that remote design teams are unable to detect can be assessed by on-site engineers who are familiar with local building requirements, environmental cycles, and regional grid behaviors.
Then there is the supply chain fragility. The longer the global supply chain, the more vulnerable and risk-sensitive it becomes. Reducing supply chains' length through local assembly and regional manufacturing reduces the likelihood of exposure to import restrictions, delays, and price volatility.
The global impact
What is happening in APAC is not a regional anomaly. It’s flagging where global infrastructure is heading.
It takes more than just implementing global solutions to succeed in APAC. As AI accelerates worldwide, regulatory complexity increases, and climate pressures intensify, other regions will face similar constraints. APAC is simply confronting them first and at scale. Bringing together global operational standards, shortened supply chains, identification of local engineering talent, and regional manufacturing can unlock scalable solutions without sacrificing reliability.
To fully understand operational reality at the site level, close integration with partners and customers is necessary. Alliances with clients and contractors that have access to workload evolution, expansion goals, and operational pain points, enable infrastructure to adjust to demand.
The signals coming from the APAC region shouldn’t just be observed from a distance. The organisations around the world that recognise and move early to rethink how they design, source, and operate infrastructure going forward won’t just navigate the pressures of AI growth, regulation and sustainability. They’ll be setting the blueprint for a resilient and scalable infrastructure worldwide.
Midge Pan is General Manager APAC at Subzero Engineering.
