Record-breaking growth in data centre project announcements is reshaping the digital infrastructure landscape across the Asia-Pacific region, with nearly 13 gigawatts ( GW ) of capacity added in the first half of 2025, a 160% year-on-year increase, according to a recent report.

The pace of expansion marks a step change in scale and ambition. Total funding requirements for this new wave of development have already surpassed US$180 billion, with hyperscalers and artificial intelligence ( AI )-focused operators at the centre of demand, finds real estate researcher Knight Frank’s latest Asia-Pacific Data Centre report.

Global technology giants are leading the charge. Collectively, AWS, Microsoft, Google and Meta have pledged more than US$160 billion in 2025 alone, underlining the intensity of infrastructure investment in support of AI and cloud workloads.

Amazon is expected to exceed US$100 billion in capital expenditure this year, up from US$82 billion in 2024, while Microsoft has already committed over US$33 billion, following US$55 billion spent last year.

“The sheer volume of new projects underscores how critical Asia-Pacific has become in the global digital infrastructure map,” notes Fred Fitzalan, Knight Frank’s head of data centres for Asia-Pacific. “Operators now face the dual challenge of rapid deployment and balancing capacity across AI and cloud use cases, all while navigating grid constraints and permitting delays.”

Key regional hubs

Johor, Malaysia, has emerged as Southeast Asia’s fastest-growing data centre hub. Aggregate capacity doubled to 5.8GW in Q2 2025, including 2.0GW in new announcements. Government-backed planning guidelines and rising AI and social media workloads drove 260 megawatts ( MW ) of take-up, leaving vacancy at a razor-thin 1.1%.

The Japanese capital Tokyo remains a core regional hub with capacity exceeding 4.2GW, though take-up slowed to 41.1MW amid constrained supply. Investment remains active, with Ares Management closing a US$2.4 billion fund via Ada Infrastructure, and Mitsui & Co. acquiring domestic assets worth US$122 million.

In Australia, meanwhile, Melbourne is becoming a key battleground, with total supply tripling to 4.7GW. Hosting all four major US cloud providers, it saw 127.6MW of take-up, 95% of which was AI-driven. Committed projects now total 934.8MW, signalling a structural shift in Australia’s data geography.

Seoul continues to attract capital, though new capacity has remained flat since Q3 2024 at 1.8GW. Take-up reached 86.2MW, largely in Q2, driven by enterprise and public cloud users. Notably, Macquarie Asset Management acquired the 40MW Hanam Data Centre for US$538.4 million, while LG U+ is expanding its AI footprint with a US$441.7 million project.

India is also growing its capacity with NTT and Blackstone–Panchshil Realty both announcing 500MW hyperscale campuses in Mumbai. The report also highlights Maharashtra’s push for green-integrated data centre parks aligned with 100% renewable energy, which aims to bring 1.5GW to market, as public cloud demand continues to dominate, accounting for 98% of H1 take-up at 97.6MW.

Investment

The funding mix is also evolving, with infrastructure and private equity investors increasingly co-developing powered shells with operators to shorten time-to-power cycles.

However, grid and permitting constraints remain widespread, particularly in tier-1 markets, an increase in data sovereignty and geopolitical considerations are also influencing project delivery timelines.

Despite these challenges, regional momentum remains robust. North Asia continues to see rising rental values as US and Chinese providers compete for the same capacity. GPU-as-a-service players are also becoming influential tenants, pushing demand for multi-megawatt, flexible-ready builds.

“The priority now is flexibility,” Fitzalan adds. “Operators must design for both AI and cloud from day one. This complexity adds cost but is now non-negotiable for winning tenants and long-term viability.”