Knight Frank’s Asia-Pacific Data Center report showed Wednesday that in the first half of this year, the region secured nearly 13 gigawatt (GW) of new project announcements, a 160 percent increase and more than double the 5GW announced in the same period last year.

The funding needed for these projects already exceeds $180 billion, the global real estate consultancy and estate agency said in a statement.

According to the statement, major technology firms are driving much of this investment.

Amazon is projected to exceed $100 billion in capital expenditure for 2025, up from about $82 billion last year, while Microsoft invested $55 billion in 2024 and has committed more than $33 billion this year.

Collectively, Microsoft, AWS, Google, and Meta have committed over $160 billion in 2025 alone, reflecting the intensity and scale of current infrastructure development.

“The sheer volume of new projects in the region highlights just how important the region has become in the global digital infrastructure landscape,

“However, coordinating this rapid growth is a complex challenge, as operators must keep pace with technological advances and rising energy needs, all while ensuring new facilities are delivered in step with evolving demands,” said Fred Fitzalan, head of data centers Asia-Pacific, Knight Frank.

Alongside the hyperscalers, graphics processing unit (GPU)-as-a-service providers are expanding rapidly, seeking multi-megawatt capacity across the region and bringing greater diversity into leasing conversations.

Creditworthiness and shortened deployment timelines remain perennial challenges, but innovative guarantee structures enable some operators to compete effectively for new contracts.

“What has become clear is the strict requirement for operators to design facilities with capacity that can be flexibly deployed for either cloud or artificial intelligence (AI) workloads, offering tenants maximum optionality,

“While this adds cost, it is now a decisive factor in site selection. Locations that combine proximity to parent sites with sufficient power allocations to support long-term runway are winning. However, this remains a significant challenge given national grid constraints and permitting delays in Tier 1 APAC markets,” Fitzalan added.

According to the statement, Johor established itself as Southeast Asia’s fastest-growing data center hub, with aggregate supply nearly doubling over the last 12 months to 5.8GW in the second quarter of 2025, including 2.0GW of new project announcements, backed by strong government support and the rollout of national Data Centre Planning Guidelines.

Johor recorded 260 megawatt (MW) of take-up in the first half of 2025, with social media accounting for 61 percent and the remainder driven by AI demand.

The market is now highly constrained, with a vacancy rate of just 1.1 percent, as planning becomes more challenging and power shortages occur.

Meanwhile, Tokyo continues to hold its position as a key regional hub with aggregate capacity exceeding 4.2GW, a 2.7 percent increase on volumes recorded at the end of the second quarter of 2024.

Investment activity remains strong, highlighted by Ares completing a $2.4 billion Japan-focused fund through Ada Infrastructure, while Mitsui & Co. Asset Management’s $122 million acquisition signals sustained domestic investment appetite.

Tokyo recorded 41.1MW of capacity transacted over the past six months.

This is a slowdown from the first half of 2024, when 286.6MW was transacted, due to reduced supply in the market.

Tokyo continues to be a tightly constrained market, with colocation vacancy rates at just 7 percent.

Melbourne, on the other hand, is stepping out of Sydney’s shadow, with total supply nearly tripling to 4.7GW as of the second quarter of 2025, as land and power constraints push development south.

The city now hosts dedicated cloud regions from all four major US providers: AWS, Microsoft, Google, and Oracle, with 95 percent of colocation take-up driven by AI workloads.

Live information technology (IT) capacity is now 337.1MW, marking a 25.4 percent year-on-year increase.

This growth trajectory is expected to continue, supported by a pipeline of 934.8MW in committed and under construction projects.

In the first half of 2025, Melbourne saw 127.6MW of transacted capacity, with AI remaining a primary driver of demand and representing 95 percent of all colocation take-up.

Seoul’s investor appetite also remains strong. LG U+ is set to grow its 87.2MW footprint with a $441.7 million investment in a new AI-focused data center.

Meanwhile, Macquarie Asset Management has acquired the 40MW Hanam Data Center for around $538.4 million.

No new capacity has been added to the market since the third quarter of 2024, keeping total supply steady at 1.8GW.

In the first half of 2025, Seoul recorded 86.2MW of leasing activity, with 85 percent of that occurring in the second quarter.

Demand was split between enterprise users and public cloud providers, alongside an emergence of Chinese tenants entering the market in a bid to diversify their AI strategies across APAC.

Mumbai is also experiencing hyperscale acceleration with NTT confirming its 500MW NAV2 campus and Blackstone partnering with Panchshil Realty on another 500MW AI facility.

The Maharashtra government’s approval of “Green Integrated Data Centre Parks” targeting 1.5GW of capacity with 100 percent renewable energy mandates signals policy alignment with industry requirements.

Over the past six months, Mumbai recorded 97.6MW of take-up in the first quarter of 2025.

Public cloud providers accounted for 98 percent of colocation demand during this period.

Knight Frank highlighted that the funding landscape is evolving, with infrastructure and private equity capital increasingly partnering with operators on developer-led powered shells to achieve faster time-to-power deployment.

Looking ahead, it noted grid capacity and power availability remain major constraints, with geopolitical considerations shaping project delivery timelines.

Despite these challenges, it said the momentum remains strong.

It noted that cloud providers in the United States and China now often compete for the same capacity, driving up rental values, particularly in North Asia.

Additionally, it said the AI ecosystem continues to expand beyond traditional hyperscale deployments.

The task ahead is to synchronize these vast expansions with technology evolution and energy demand, building digital infrastructure that is both flexible and future-ready, it added.

APDCA foresees $34 bln economic output from Malaysia’s data centers