2026年亚太数据中心趋势与展望

2026 Asia PacificData CentreTrends & Outlook REPORT 2026Asia Pacific Data CentreTrends & Outlook AI demanddrives newdata centredevelopment Big Techplans toincreaseCapExby 61% in2026tosupport AI businessgrowth Global hyperscale demand isgrowing ata CAGR of 14% Neoclouddemand continuesto emerge rapidly Some new developments areleveraging modularconstruction to expeditedevelopment timeframes Power availability remainsthe biggest challenge Datacentreconstructioncosts are escalating; Tokyo isthe most expensive market Data centre expansioncontinues to outpace utilitycapacity Upcoming data centreprojects shift locational focus Australiancitiesand non-corelocations in Japanhold strongpotential Malaysia and Indiaareemergingas leadinghyperscale hotspots Emerging markets such asThailand and Indonesia alsooffer attractive opportunities Different routes available fordata centre investment Direct investment volumereaches arecord high US$11.6billion in 2025 Datacentreyields widenalong with interest rate hikesin Australia and Japan Investors alsoconsiderplatform &equity investments Contents AI demanddrivesnewdata centredevelopmentPower availability remains the biggest challengeUpcomingdata centreprojects shift locational focus010203 Different routesavailable for data centre investment04 Recommendations05 Market Profiles06 01AI demand drives new datacentre development DEMAND DRIVERS Growth in AI demand gathers pace The global datacentremarket remains dominated by cloud andcorporate usage, with AI contributing less than 15% of totalworkload as of 2025. With most AI demand now concentrated inthe U.S., the ratio in Asia Pacific is even smaller. However, as more capital flows into AI development and more AIuse cases and adoption occurs, demand for both AI training andAI inference is increasing. Goldman Sachs data show that the four-year compound annualgrowth rate (CAGR) (2023–2026) of global AI demand is 98%. Global hyperscale demand is growing at a CAGR of 14% whilenon-hyperscale demand is growing at 4%, showing that AIdemand is far outpacing non-AI usage. At the infrastructure level, AI workloads require substantiallyhigher power capacity, greater rack density, and more advancedbuilding specifications than conventional datacentreworkloads. This is driving the development of new design standards,particularly in cooling systems, which are being rapidly adopted inboth greenfield developments and the retrofitting of suitableexisting datacentres. DEMAND DRIVERS Big tech drives AI infrastructurebuild-out CapExon AI is experiencing unprecedented growth as tech firms invest in necessary, massive infrastructure to secure a dominantposition in the AI race. U.S.hyperscalersremain among the most aggressive players,shelling out vast sums on cloud infrastructure, AI training, and AIinference. Amazon has announced that its 2026 AI spend will reach US$200billion, while Google plans investment of US$180 billion. As well as investing in their home market, U.S.hyperscalersareexpanding across Asia Pacific, deploying datacentresfor cloudand AI usage. Mainland Chinese tech companies are also stepping up theirCapExin AI,abeitat a lesser magnitude. ByteDance has earmarked investment of around US$23 billion in2026, with Tencent and Alibaba planning a smaller outlay. DEMAND DRIVERS Power-rich markets lead regional growth Figure 3: Q1 2026 Asia Pacific Data Centre Live Capacity and Growth byMarket Asia Pacific datacentredemand continued to expand at pace in2025, underpinned byongoing cloud expansion and the rapidscaling of AI-related workloads.New sources of demand are emerging along the AI developmentvalue chain, with a growing number ofNeocloudoperators seekingrapid expansion across the region to support GPU-intensive AI In terms of live capacity growth, Johor led the region in 2025 withtraining and inference workloads. This is adding depth and diversityto the demand base beyond traditional hyperscale cloud providers. growth of 53% y-o-y, followed by Melbourne with expansion of Government-led AI initiatives are also providing tailwinds todemand, with Australia’s National AI Plan and Korea’s “AI for All”scheme among several similarframeworks introduced across theregion. These initiatives are expected to drive long term demandand support the development of local AI ecosystems by Whilemature markets such as Singapore and Hong Kong SARregistered more modest growth, this was due todevelopmentconstraints as opposed to a lack of demand. In mainland China,encouraging investment, stimulating R&D activity, and acceleratingthe build-out of AI-ready digital infrastructure. growth slowed as vacant space continued to be absorbed.These dynamics reflectthe growing concentration of demand inpower-advantaged locations capable of supporting large-scalecloud and AI deployments.Malaysia, Australia and India aresecuringa larger share of incremental demand ashyperscalersprioritisescalability and delivery certainty in the ra