> **来源：[研报客](https://pc.yanbaoke.cn)** ```markdown # ASIA PACIFIC DATA CENTRE CONSTRUCTION COST GUIDE 2025 Summary ## Core Content Overview This report from Cushman & Wakefield provides an in-depth analysis of data centre construction and land costs across the Asia Pacific region, highlighting key themes shaping the sector. It includes insights on power availability, the impact of AI, and investment trends, offering a comprehensive guide for operators, developers, and investors. ## Main Themes and Key Points ### 1. Power Availability & Cost - **Power as a Critical Factor**: Power is a key operational cost and challenge for data centres in the region. Most markets face long wait times for new power connections, ranging from 18 to 36 months, with some locations experiencing delays of up to 5–7 years. - **Power Tariffs and Costs**: - The average power tariff in Asia Pacific in November 2024 was 11.48 cents/kWh. - Power costs vary significantly across markets, with Singapore being one of the most expensive at 19.19 cents/kWh. - A 0.1 reduction in PUE (Power Usage Effectiveness) can save operators up to \$1.9 million annually. - **Power Scarcity**: Powered greenfield sites are becoming increasingly scarce, pushing land prices higher and encouraging brownfield redevelopment. - **Power Sources**: - Renewable power is growing in demand due to cost, regulation, and environmental concerns. - Some markets, like Japan, are considering nuclear power as an alternative. - **Regional Variations**: - **India**: Second-largest renewable power capacity in APAC, but 80% of power still comes from fossil fuels. - **Hong Kong**: Has the lowest power tariff growth, with some recent investments in green energy. - **Singapore**: Faces strict regulations and high tariffs, making it a constrained market. - **Japan**: Tokyo's power shortage is limiting current development, with operators exploring alternative locations like Tsukuba and Osaka. - **South Korea**: Seoul has affordable power but lacks suitable land. - **Philippines**: Low power capacity and frequent outages are issues, with Manila's tariff nearly double that of some Indian and Malaysian cities. - **Vietnam**: Struggles with power grid capacity, but the government is working to add ~6.6 GW of new capacity. - **Thailand**: Power availability is improving, with some clusters having faster connection times. ### 2. The Impact of AI - **AI Development Pipeline**: While the full impact of AI on the data centre sector is expected in several years, early signs of planning and investment are visible. - **Location Strategy**: AI data centres are likely to be built near existing infrastructure and talent pools to avoid cannibalizing power availability in existing zones. - **Design Adaptations**: Many data centres are being redeveloped into hybrid models that can support both cloud and AI workloads, with increased cooling requirements. - **Cooling Technologies**: Liquid cooling is being explored, but adoption is still in early stages due to concerns over durability, maintenance, and leakage. - **Data Sovereignty Challenges**: New laws may restrict AI data centres to a single market, requiring operators to comply with varying data regulations. ### 3. Investment Trends - **Development Pipeline**: Over 13.5 GW of colocation capacity is under construction or in planning across the region, with about 32% concentrated in Sydney, Tokyo, Singapore, Hong Kong, and Mumbai. - **Capital Requirements**: The development pipeline requires over \$125 billion in capital and offers the potential to generate \$14 billion in annual colocation rent. - **Yield-on-Cost**: Developers can expect yield-on-cost ratios of 13% to 16%, with top 5 cities seeing higher returns (15.6%). - **Consolidation**: The sector is seeing a shift towards larger-scale investments, often involving consortiums, as smaller plays become less attractive. - **Transaction Activity**: Legacy data centres in Japan, South Korea, and Singapore are seeing increased transaction activity, reflecting a more mature market. ## Key Markets and Costs | Market | Power Tariff (USD cents/kWh) | Annual Power Cost (Millions) | Monthly Power Cost ($/kW) | |----------------|-----------------------------|-----------------------------|--------------------------| | Singapore | 19.19 | 50.4 | 138 | | New Zealand | 14.94 | 39.3 | 108 | | Japan (Tokyo) | 14.22 | 37.4 | 102 | | Taiwan (Taipei) | 13.81 | 36.3 | 99 | | Philippines | 13.59 | 35.7 | 98 | | Indonesia | 12.23 | 32.1 | 88 | | Australia | 11.97 | 31.5 | 86 | | Thailand | 10.48 | 27.5 | 75 | | China (Shanghai)| 10.09 | 26.5 | 73 | | Hong Kong | 9.82 | 25.8 | 71 | | South Korea | 9.14 | 24.0 | 66 | | Malaysia | 7.32 | 19.2 | 53 | | India (Mumbai) | 6.71 | 17.6 | 48 | | Vietnam (HCMC) | 7.21 | 18.9 | 52 | ## Key Assumptions and Specifications ### Base Build | Category | Low Specification | Mid Specification | High Specification | |-----------------------------|-------------------|------------------|--------------------| | Quality of Construction | Average quality | Good quality | Prestige quality | | Redundancy & Operational Standard | Tier 2 | Tier 3 | Tier 3-4 | | Power Source | 1 path from 1 substation | Multiple paths (1–2 substations) | Multiple paths, multiple substations | | Cooling Technology | Air Cooled | Air Cooled | Liquid Cooling | ### Fit Out | Category | Low Specification | Mid Specification | High Specification | |-----------------------------|-------------------|------------------|--------------------| | Quality of Construction | Average quality | Good quality | Prestige quality | | Design for Maintenance | No | Yes | Yes | | Compartmentalisation | No | Yes | Yes | | Space | Cabinets | Cabinets / Cages | Cabinets / Cages / Private Suites | | Services | Low Density | Mid Density | High Density / Private AI | | Floor Loading | Low Capacity | Mid-High Capacity | Mid-High Capacity | ### Land | Category | Low Range | Mid Range | High Range | |-----------------------------|-----------|----------|-----------| | Land Price Quality | Average quality | Good quality | Prestige quality | | Location | Remote, Low Demand, Minimal Infrastructure | Suburban, Moderate Demand, Infrastructure Available | Prime Location, High Demand, Infrastructure Available | ## Regional Land Cost Index - **Top Markets**: The 18 most expensive land cost clusters are found in mature markets, with Central Jakarta as the fourth most expensive cluster due to high demand. - **Inner City Dominance**: Inner city clusters dominate the most expensive rankings, but transaction volumes are limited. - **Methodology**: Land cost rankings are based on clusters rather than cities to avoid distortion from inner-city colocation and retail clusters. ## Conclusion - The Asia Pacific data centre market continues to grow, driven by digitization and AI demand. - Power availability and cost remain the primary challenges, with operators exploring alternative solutions and locations. - Investment trends are shifting towards larger-scale developments and consortiums. - The sector is becoming more mature, with increased focus on sustainability and operational efficiency. ```