This article was first published by Frontier Enterprise on 6 February 2026. 

Most data centre conversations in Singapore still focus on the wrong question. After two decades delivering critical infrastructure across Southeast Asia’s constrained markets, how you build now matters more than what you build. 

AI has changed the equation. What were once manageable delays in data centre delivery have become decisive disadvantages as AI-driven growth accelerates across the region. AI workloads are often committed upfront to fixed timelines, leaving little tolerance for delivery slippage. Where planning, power and delivery are aligned early, projects move from approval to operation. Where they are not, opportunity quickly shifts elsewhere. 

The issue isn’t capital or demand. It’s execution, and not all markets are equally prepared. 

Why traditional playbooks fail here 

Here’s what many operators learn quickly in Southeast Asia. Delivery models that work in Frankfurt or Virginia do not always translate cleanly into markets like Jakarta or Bangkok. I’ve seen multinational teams run into difficulty when they apply standardised specifications without adapting to local power, regulatory and construction realities. 

By contrast, regional players who adapt execution to local conditions often move faster, even with simpler specifications. 

Singapore offers a telling reference point. Severe land and sustainability requirements leave little margin for error. This forces tighter coordination between policy, power planning, and project delivery from the outset. In markets with fewer constraints, that discipline is easier to defer, allowing approvals, infrastructure readiness, and construction timelines to drift out of alignment. 

The AI acceleration that changes everything 

I’ve observed AI workloads rewriting the rules of delivery economics. Unlike traditional demand, AI compute is scarce and deployments are tied to fixed training and launch windows, leaving little room for delay. 

That pressure is already visible across the region. AI’s demand for processing power is driving a surge of investment across Southeast Asia. Global technology firms and hyperscalers are committing billions of dollars to new data centre capacity in markets such as Malaysia and Thailand. This is creating a more competitive, multi-node “AI archipelago.” 

Yet analysts caution that despite rapid capacity growth, supply could still lag demand by 15-25 GW by 2028. Limited AI-ready infrastructure and access to power remain key constraints. 

The result is uneven readiness. When power, permits, or essential utilities lag, delays don’t just shift timelines; they shift where AI workloads end up. 

Making regional complexity manageable 

Across large infrastructure and data centre projects in Southeast Asia, execution risk rarely stems from a single technical constraint. More often, it emerges at the interfaces where power readiness, regulatory approvals, procurement and construction sequencing intersect. These interfaces cut across government agencies, utilities, industry players and technology providers. 

What separates projects that progress from those that stall is not the sophistication of individual solutions. It’s how early and effectively these interfaces are managed. Decisions around construction methods and the use of more eco-friendly, sustainable building materials increasingly influence both approvals and long-term performance. 

When delivery governance is fragmented, risks surface late and timelines narrow quickly. This pattern is not unique to Asia: In the Middle East’s giga-scale data centre build-out, rapid expansion and compressed timelines have intensified pressure on delivery interfaces and contractor capacity. 

At this scale, delivery disciplines matter. Integrated project controls, robust risk frameworks and proactive interface planning become decisive in preventing early bottlenecks from stalling capacity. 

What must happen now 

For regional operators, realism matters. Markets can no longer be assessed on announced capacity or incentives alone. Where execution conditions such as power readiness, permitting and contractor availability are underestimated, delays quickly erode returns. Delivery feasibility now needs to be tested as rigorously as site economics. 

For policymakers, capital attraction is only part of the equation. Execution ecosystems and delivery governance matter just as much. This includes skilled contractors, predictable approvals, clear sustainability requirements and early coordination with utilities. Where these lag, investment momentum becomes harder to sustain, regardless of incentives. 

For investors, execution risk deserves sharper scrutiny. Differences in delivery readiness increasingly translate into differences in timing, certainty and realised returns. Pricing those risks more carefully, rather than assuming uniform outcomes, will be critical as AI-driven demand accelerates. 

The next decade’s winners 

Delivery capability has become a differentiator. As AI compresses timelines and removes slack from infrastructure planning, markets that can consistently align planning, power, and delivery are already pulling ahead. 

Over the next decade, that gap will widen. In Singapore and across the region, markets that can turn approvals into operational infrastructure will consolidate their position. Those that cannot will find that ambition alone no longer carries weight.