Key Takeaways

  • 15 new data centre facilities were announced across ASEAN in 2026, the highest volume in a single year
  • Malaysia has emerged as ASEAN's fastest-growing data centre market, driven by lower land costs and power availability
  • Singapore's data centre moratorium was lifted in 2022 but sustainable power constraints still limit new builds
  • AI GPU workloads are driving demand for higher-power-density facilities — 10kW+ per rack versus traditional 3-5kW
  • Total data centre investment announced for ASEAN in 2026 exceeds $20 billion

The Facts

ASEAN is experiencing an unprecedented data centre construction boom, with 15 new facility announcements in 2026 representing the highest volume in a single calendar year. Total investment announced exceeds $20 billion, driven by the convergence of cloud infrastructure expansion from hyperscalers, data sovereignty requirements pushing workloads to in-country infrastructure, and the AI GPU compute demand that is the dominant new workload category for data centre capacity planning.

Malaysia has emerged as ASEAN's fastest-growing data centre market — surpassing Singapore in new facility announcements for the first time. Johor state, bordering Singapore, has become particularly active: lower land costs, power availability, and Singapore's proximity (enabling Singaporean enterprises to benefit from lower-cost Malaysian infrastructure while maintaining close geographic connectivity) create a compelling combination for both hyperscalers and colocation providers.

Microsoft committed $2.2 billion to Malaysia data centre expansion; Google committed $2 billion to Thailand; AWS expanded across Singapore, Malaysia, and Indonesia. These hyperscaler commitments are supplemented by regional colocation providers — Equinix, Bridge Data Centres, and local operators — expanding capacity to meet demand from enterprises that prefer multi-cloud or cloud-neutral infrastructure.

Technical Deep-Dive

AI GPU workloads are fundamentally different from traditional cloud compute in their power and cooling requirements. Traditional server racks are designed for 3-5kW per rack; H100 GPU clusters typically require 10-20kW per rack, with next-generation accelerators pushing higher. This power density difference means that facilities built for traditional cloud workloads cannot simply host AI GPU clusters — they require infrastructure upgrades or replacement.

New ASEAN data centres announced in 2026 are being designed from the ground up for AI workload densities, with higher-power electrical infrastructure, liquid cooling capabilities (air cooling is insufficient for the highest-density GPU racks), and direct utility connections that bypass the power constraints of urban data centre campuses.

Network connectivity between AI GPU clusters and storage infrastructure is the other technical dimension. AI training requires moving massive datasets from storage to GPU memory at high bandwidth — facilities with insufficient internal networking create training throughput bottlenecks that waste expensive GPU time.

The ASEAN Perspective

Malaysia's emergence as a data centre hub creates economic opportunities that extend beyond the construction phase. Data centres employ skilled operations staff, generate demand for technical maintenance services, and anchor broader technology cluster development around the facilities.

Singapore benefits indirectly from Malaysia's data centre expansion: Singapore-based enterprises can access lower-cost infrastructure in Johor while maintaining Singapore headquarters operations and regulatory compliance, with high-bandwidth connectivity between the two markets (the Singapore-Malaysia subsea cable infrastructure is world-class).

For ASEAN enterprises planning cloud and AI infrastructure strategy, the data centre boom creates a more competitive supply environment than existed 18 months ago. New capacity from multiple providers in multiple ASEAN markets gives enterprise buyers more negotiating leverage and more options for achieving data residency requirements at competitive pricing.

RECATOOLS Verdict

ASEAN's data centre boom is a direct consequence of AI demand — and it is well-founded. The compute requirements for AI model training and large-scale inference are genuinely large, and the regional demand for AI services will continue to grow as enterprise AI adoption accelerates.

The strategic question for ASEAN governments is ensuring that the infrastructure investment benefits local economies beyond construction jobs — developing the skills, supply chains, and secondary industries that transform data centre infrastructure into durable economic capability.

Frequently Asked Questions